The oligomeric state of CtBP determines its role as a transcriptional co-activator and co-repressor of Wingless targets

Monomeric CtBP activates Wg signalling in flies

CtBP is thought to exist in an equilibrium between monomeric (Kim et al, 2005; Zhao et al, 2009), homo-dimeric and possibly higher ordered homo-oligomeric complexes (Kumar et al, 2002; Balasubramanian et al, 2003; Nardini et al, 2003; Shi et al, 2003; Thio et al, 2004; Kim et al, 2005; Mani-Telang et al, 2007; Kuppuswamy et al, 2008; Zhao et al, 2009). While the native oligomeric state has mostly been determined for mammalian CtBP proteins, the entire dehydrogenase domain of fly CtBP is highly conserved (e.g. fly CtBP and hCtBP1 domains are 72% identical with 84% similarity). Nearly all of the residues making inter-molecular contact in the hCtBP1 homo-dimers are identical in fly CtBP (Kumar et al, 2002). This information was utilized to construct a fly CtBP protein that should not be able to dimerize, and thus remain monomeric.

There are several different isoforms of fly CtBP predicted to express proteins containing 383, 386, 476 and 479 residues (Poortinga et al, 1998; Nibu et al, 1998b; Sutrias-Grau and Arnosti, 2004). The short and long isoforms differ in their C-termini, downstream of the dehydrogenase domain. A minigene expressing a CtBP short isoform under the control of its endogenous regulatory elements can complement the CtBP mutant phenotype (Zhang and Arnosti, 2011). We have previously shown that both short and long isoforms can activate Wg/Arm-dependent transcription (Fang et al, 2006). Hence, a short isoform (383 aa) was used for all subsequent experiments in this report.

To generate a monomeric CtBP, four conserved residues, previously shown to be important for hCtBP1 self-association (Kumar et al, 2002), were mutated in fly CtBP. The resulting CtBP variant is referred to as CtBP^Mono. The C134Y and N138R substitutions should result in steric and electrostatic hindrance, hence preventing homo-dimerization and the R141A and R142A mutations should disrupt inter-molecular salt bridges and hydrogen bonds as predicted for hCtBP1 (Kumar et al, 2002). The normal equilibrium between monomers and dimers in wild-type CtBP (CtBP^WT) should be dramatically shifted to the monomeric state for CtBP^Mono (Figure 1A).

To test if the mutations in CtBP^Mono abolished its ability to self-associate, differentially tagged CtBP forms were co-transfected in the Drosophila hemocyte-derived cell line Kc167 (Kc) and assayed for binding using co-immunoprecipitation (co-IP). While CtBP^WT-Flag could co-IP CtBP^WT-HA, it was not able to pull-down CtBP^Mono-HA (Figure 1B). Mutations also disrupted the ability of CtBP^Mono to homo-oligomerize as judged by this assay (Figure 1B). These results demonstrate that CtBP^Mono cannot dimerize, resulting in a ‘forced monomer’ version of CtBP.

To test the role of dimerization in regulating Wg signalling, transgenic lines were constructed containing CtBP^WT or CtBP^Mono under the control of the Gal4/upstream activating sequence (UAS) inducible promoter. These CtBPs can then be expressed in any fly tissue for which a Gal4 driver line is available (Phelps and Brand, 1998). We previously reported that overexpression of CtBP, via nearby insertion of a P[GSV] element (Toba et al, 1999) could suppress a small eye phenotype caused by overstimulation of Wg signalling in the developing eye (Fang et al, 2006). Consistent with these results, several P[UAS-CtBP^WT] lines were able to suppress the small eye phenotype caused by GMR-Gal4-dependent expression of a stabilized form of Arm (Arm^*) (Figure 2A and B). Thus, CtBP antagonizes Wg signalling downstream of Arm stabilization in this assay. In stark contrast, misexpression of CtBP^Mono caused a significant enhancement of the GMR-Gal4::UAS-arm* small eye phenotype (Figure 2A and C), suggesting that CtBP monomers promote Wg signalling in this context.

Wg signalling also has a significant role in defining the wing margin that originates from the dorsal/ventral (D/V) boundary of the wing imaginal disc. Antagonism of Wg signalling in this tissue leads to a loss of the wing margin, causing notches in the adult wing (Phillips and Whittle, 1993; Couso et al, 1994). To assay the role of CtBP^Mono in Wg-directed wing margin formation, a sensitized genetic background was created by misexpression of Pygo at the anterior/posterior (A/P) boundary of the wing disc using Patched-Gal4 (Ptc-Gal4) (Figure 2D–F). Although Pygo is known to positively regulate Wg signalling (Belenkaya et al, 2002; Kramps et al, 2002; Parker et al, 2002; Thompson et al, 2002), misexpression of Pygo antagonizes Wg signalling, possibly due to disruption of the stoichiometry of a protein complex (Parker et al, 2002).

In the Ptc-Gal4::UAS-Pygo background employed in this assay, over 80% of the adult wings displayed notches. These notches were categorized into two groups based on their size. Small notches had loss of wing margin only between the L3 and L4 veins (see Figure 2E). Big notches extended beyond these veins (see Figure 2F). Co-expression of CtBP^Mono significantly suppressed the loss of wing margin caused by Pygo, with a dramatic reduction in the frequency of big notches (Figure 2G). These data provide another line of evidence supporting a positive role for CtBP^Mono in the regulation of Wg signalling.

To further test the role of CtBP^Mono in Wg signalling, expression of a Distalless enhancer trap line (Dll-lacZ) was monitored. In larval third instar wing imaginal discs, Dll-lacZ is activated by Wg in a broad domain centred on the D/V boundary of the presumptive wing blade (Zecca et al, 1996; Neumann and Cohen, 1997) (Figure 3A). Transgenic flies carrying UAS-CtBP^WT or UAS-CtBP^Mono transgenes were crossed to a Engrailed-Gal4 (En-Gal4) driver, leading to expression of transgenes in the posterior half of the disc (Figure 3E and H). Lines expressing CtBP^WT and CtBP^Mono at similar levels resulted in an enhancement in the Dll-lacZ expression (Figure 3D, F, G and I; Supplementary Figure S1). These results provide additional support for positive regulation of the Wg pathway by CtBP^Mono.

CtBP monomers promote activation of Wg targets in Kc cells and CtBP acts downstream of Pygo in activating transcription

We have previously shown that the expression of the genes CG6234 and naked cuticle (nkd) is activated by Wg signalling in Kc cells (Fang et al, 2006). In the absence of signalling, CtBP and TCF act in parallel to repress nkd expression, but CtBP is not required for activation of nkd expression by Wg signalling (Fang et al, 2006). In contrast, CtBP repression of CG6234 in the absence of signalling is minimal, but CtBP is required for maximal activation of CG6234 upon Wg stimulation (Fang et al, 2006).

To test whether the positive regulation of CG6234 by CtBP is occurring at the transcriptional level, a reporter gene containing a minimal WRE from this target gene was examined. Figure 4A shows the location of a minimal WRE (539 bp) derived from a previously reported 2.2 kb WRE (Fang et al, 2006; see Materials and methods), which is comparatively more responsive to Wg signalling (data not shown).

The CG6234 WRE reporter was highly activated by expression of Arm^* in a TCF-dependent manner (Figure 4B). RNAi-mediated depletion of CtBP also caused a dramatic reduction in activation of the CG6234 WRE reporter (Figure 4B). Consistent with the data for CG6234 transcripts (Fang et al, 2006), there was still residual activation of the CG6234 WRE reporter by Arm^* in CtBP RNAi-treated cells (Figure 4B). These data suggest that CtBP acts in parallel with other co-activators, which are recruited by Arm for activation of CG6234 WRE reporter.

To test the role of CtBP dimerization in promoting Wg-mediated transcriptional activation of the CG6234 WRE reporter, a CtBP gene replacement strategy was employed. Endogenous CtBP was depleted using dsRNA corresponding to the 5′UTR of CtBP. These cells were then transfected with the reporter, plus transgenes expressing CtBP^WT or CtBP^Mono. These CtBP transgenes contained a heterologous 5′ UTR, so they were not targeted by the CtBP RNAi. Activation of the CG6234 WRE reporter by Wg-conditioned media was then assayed. Transfection of CtBP^Mono rescued the CtBP RNAi defect to a similar level as seen with CtBP^WT transfection (Figure 4C; compare the second and third groups). A similar rescue of CG6234 WRE reporter activation by CtBP^Mono was observed when the Wg pathway was stimulated by expression of Arm^* (Figure 4D). CtBP^Mono was expressed at similar levels as CtBP^WT in these experiments (Figure 4D and data not shown). These results demonstrate that CtBP^Mono is capable of substituting for endogenous CtBP to promote activation of the CG6234 WRE reporter.

CtBP has previously been shown to be required for activation of Gal4–Arm^*-dependent activation of a UAS-luc reporter (Fang et al, 2006). Both CtBP^WT and CtBP^Mono had no effect on UAS-luc when co-expressed with Gal4DBD (Gal4 DNA-binding domain). However, both CtBP forms dramatically enhanced the ability of Gal4–Arm^* to activate UAS-luc (Figure 4E). Taken together with the data from Figure 4D, these results indicate that like CtBP^WT, CtBP^Mono is functioning downstream of Arm to activate Wg transcriptional targets.

Arm contains at least two domains that contribute to activation of gene expression, a N-terminal and a C-terminal transactivation domain (Hecht et al, 1999; Stadeli and Basler, 2005; Fang et al, 2006). CtBP is recruited by the N-terminal domain of Arm for transcriptional activation (Fang et al, 2006). The N-terminal domain of Arm binds to Lgs, which serves as an adaptor to recruit Pygo to the TCF–Arm complex (Kramps et al, 2002). In order to explore a possible connection between CtBP and Pygo in activating transcription, we tested the ability of a Gal4–Pygo fusion protein to activate the UAS-luc reporter when endogenous CtBP was depleted. The activation of UAS-luc reporter by Gal4–Pygo was markedly reduced in cells treated with CtBP RNAi (Figure 4F). Gal4–Pygo has previously been shown to activate transcription independently of Arm and Lgs (Stadeli and Basler, 2005), suggesting that CtBP acts downstream of Pygo to activate Wg target gene expression.

CtBP dimerization is required for its antagonistic role in Wg signalling in Kc cells

We have previously identified three WREs in the nkd locus, two upstream of the nkd transcriptional start site (nkd-UpE1 and nkd-UpE2) and one in the first intron of nkd (nkd-IntE) (Chang et al, 2008). While reporters for all three WREs were derepressed upon depletion of CtBP, the nkd-UpE1 reporter consistently exhibited the largest response (Supplementary Figure S2). In addition, TCF knockdown also caused derepression of nkd-UpE1, while having no effect on nkd-UpE2 or nkd-IntE (Supplementary Figure S2). Therefore, nkd-UpE1 (Figure 5A) was used for all subsequent experiments to assay the role of CtBP oligomers in regulating this nkd WRE. Similar to nkd mRNA (Fang et al, 2006), simultaneous knockdown of CtBP and TCF resulted in a far greater derepression of nkd-UpE1 reporter activity than with either factor alone (Figure 5B). TCF has already been shown to be enriched at the UpE1 WRE (Chang et al, 2008). As judged by chromatin IP (ChIP), CtBP was also enriched at the nkd-UpE1 (Supplementary Figure S3), supporting a direct role for CtBP in repression of this WRE.

To test if dimerization of CtBP was required for inhibition of Wg targets in the absence of signalling, the ability of CtBP^Mono to repress the nkd-UpE1 reporter was assayed. This was done using a similar gene replacement strategy as described in Figure 4. As expected, transiently expressed CtBP^WT was able to repress the nkd-UpE1 in the absence of signalling. Strikingly, CtBP^Mono was unable to perform this function (Figure 5C). CtBP^Mono sometimes caused greater derepression of nkd-UpE1 than the control (Figure 5C), possibly due to a weak dominant negative effect on CtBP repressive activity, but this effect was not always observed (see Figure 6D). Mutations that abolish self-association of CtBP, while having no affect on its ability to promote Wg signalling, severely disrupt its ability to repress Wg target gene expression in the absence of signalling.

The approach described above is similar to that used in several other studies to provide evidence that CtBP dimerization is required for transcriptional repression (Kumar et al, 2002; Kuppuswamy et al, 2008; Zhao et al, 2009), that is, correlating loss of dimerization with loss of CtBP activity. However, this approach cannot rule out that the mutations disrupting homo-dimerization also affect other aspects of CtBP function. To provide a more convincing demonstration of the importance of CtBP self-association in antagonizing Wg signalling, monomeric versions of CtBP were created that cannot homo-dimerize, yet possess the ability to hetero-dimerize with each other. If dimerization is essential for repression by CtBP, then the monomeric forms should not be able to repress Wg targets but co-expression of these complementary monomeric forms should reconstitute dimerization and hence the repressive function of CtBP.

The strategy for engineering complementary monomeric forms of CtBP required identifying the salt bridges in the CtBP dimer and then switching the positions of the acidic and basic residues forming the salt bridge. Such inter-molecular salt bridge swaps have been previously used to show interaction or self-association of various proteins (Xiao et al, 1999; Watt et al, 2001; Venkatachalan and Czajkowski, 2008). Using the structural information of the highly conserved hCtBP1, two salt bridges (E126–R173 and E127–R171) formed by conserved residues were targeted for a swap. CtBP^Basic contains E126R and E127R substitutions, while CtBP^Acidic has R173E and R171E alterations. When expressed on their own, CtBP^Basic and CtBP^Acidic should be monomeric but have the ability to form CtBP^Basic/CtBP^Acidic hetero-dimers when co-expressed (Figure 6A).

As expected, co-expression of CtBP^Basic and CtBP^Acidic led to association of these molecules at levels similar to those seen with CtBP^WT (Supplementary Figure S4; Figure 6B, lane 2). Co-expression of differentially tagged versions of CtBP^Acidic did not result in an appreciable co-IP (Figure 6B, lane 1). Assaying self-association of CtBP^Basic was complicated by the fact that the V5-tagged version of this protein was somewhat unstable when expressed with a Flag-tagged CtBP^Basic (Figure 6C, lane 1). Stability was greatly increased by co-expression with CtBP^Acidic (Figure 6C, lane 2). Although V5-tagged CtBP^Acidic was more readily expressed, it also appeared to be more stable in the presence of the complementary CtBP^Basic (Figure 6C, lanes 3 and 4). In contrast to the V5-tagged proteins, the Flag-tagged versions were relatively stable when expressed under all conditions (Figure 6C). Taken together, these data demonstrate that the CtBP^Acidic and CtBP^Basic mutants function as predicted, being unable to homo-dimerize but capable of efficient hetero-dimerization.

When tested for their ability to rescue the derepression of the nkd-UpE1 reporter in cells depleted of endogenous CtBP, neither CtBP^Acidic nor CtBP^Basic were able to provide significant repressive activity, similar to the original CtBP^Mono mutant (Figure 6D). Remarkably, co-expression of CtBP^Basic and CtBP^Acidic restored the inhibition of nkd-UPE1 in the absence of signalling, to a similar extent as observed with CtBP^WT. These data provide compelling evidence that self-association is required for the function of CtBP as a co-repressor of a Wg transcriptional target.

CtBP dimerization is required for its ability to inhibit wing margin formation

To extend the findings summarized in Figures 5 and 6 to fly tissues, the ability of CtBP^WT and CtBP^Mono to affect development of the wing was examined. Expression of CtBP^WT at the A/P boundary of the wing disc using the Decapentaplegic-Gal4 (Dpp-Gal4) driver gave rise to a high frequency of wing notches (Table I). Expression of CtBP had no effect on the expression of Wg (Supplementary Figure S5). This phenotype is consistent with a reduction in Wg signalling at the presumptive wing margin. In contrast, expression of CtBP^Mono never resulted in wing notching (Table I). These data provide further support for a model where CtBP self-association is required for the ability of CtBP to antagonize events downstream of Wg expression in the wing primordium.

To confirm that CtBP dimerization was required for inhibition of wing margin formation, CtBP^Basic and CtBP^Acidic transgenes were also tested for a wing phenotype when misexpressed using Dpp-Gal4 (Table I). Flies containing two UAS transgenes each were generated in the following combinations: Acidic/Acidic, Basic/Basic or Acidic/Basic. While expression of both complementary combinations (Acidic¹/Basic¹ or Acidic²/Basic²) resulted in significant wing notching, Acidic¹/Acidic² or Basic¹/Basic² combinations did not (Table I). Immunostaining with CtBP antisera was performed to ensure that comparisons were made with CtBP variant proteins expressed at similar levels (Supplementary Figure S5). The CtBP Acidic¹/Basic¹ and Acidic¹/Acidic² backgrounds were expressed at similar levels, while the Basic¹/Basic² and Acidic²/Basic² combinations were expressed at slightly lower levels (Supplementary Figure S5). As with CtBP^WT and CtBP^Mono, Wg expression at the presumptive margin was not affected by any of the Acidic/Basic combinations (Supplementary Figure S5). These results indicate that dimerization of CtBP is required for antagonism of Wg signalling during wing margin formation.

A monomeric pool of CtBP in Kc cells

To assess the distribution of the monomeric and oligomeric pool of CtBP in cultured Kc cells, the association of V5-tagged CtBP^WT (WT V5) and Flag-tagged CtBP^WT (WT Flag) was examined. Excess WT Flag was expressed, in order to drive most of the WT V5 into a heteromeric complex (Figure 7A, lanes 9–12). As expected, WT V5 was found to associate with WT Flag (Figure 7A, lanes 5–8). In addition, there was a considerable amount of WT V5 present in the FLAG immunodepleted supernatants, even when a 10-fold higher level of WT Flag was expressed (Figure 7A, lane 4). This suggests the existence of a substantial monomeric pool of CtBP.

In order to exclude the possibility that the non-precipitated WT V5 CtBP was in a homo-oligomeric state, a similar immunodepletion was performed using V5-tagged CtBP^Acidic (Acidic V5) and Flag-tagged version of CtBP^Basic (Basic Flag) (Figure 7B). These CtBP mutants are unable to form homo-oligomers (Figure 6B). An increasing dose of Basic Flag was expressed (Figure 7B, lanes 9–12), and found to associate with Acidic V5 (Figure 7B, lanes 5–8). As was found with WT CtBPs, there was a substantial amount of Acidic V5 in the immunodepleted supernatants, even when 10-fold higher Basic Flag was expressed (Figure 7B, lane 4). These data support the view that a significant pool of CtBP is present as monomers in Kc cells.

The oligomeric state of CtBP determines its effect on Wg signalling

CtBP is well known for its role as a co-repressor for many transcription factors (Turner and Crossley, 2001; Chinnadurai, 2007; Kuppuswamy et al, 2008). It is also known to antagonize Wnt/β-cat signalling, possibly by binding to some TCFs (Brannon et al, 1999; Valenta et al, 2003) or by acting with APC to divert β-catenin away from TCF (Hamada and Bienz, 2004; Sierra et al, 2006). In Drosophila Kc cells, we have previously shown that CtBP works in parallel with TCF to repress expression of the Wg target nkd in the absence of signalling (Fang et al, 2006). In addition, we found that CtBP was required for activation of several Wg targets in cultured cells and fly tissues (Fang et al, 2006). Our data indicate that CtBP can both repress and activate the Wg pathway in a gene-specific manner.

In this report, we provide a dramatic example of this differential regulation of Wg-mediated transcription by CtBP using WRE reporter constructs. While CtBP is required for silencing the nkd-UpE1 reporter in the absence of signalling (Figures 5 and 6), depletion of CtBP results in a significant reduction of the CG6234 WRE reporter activation upon Wg stimulation (Figure 4). Since both these WREs are directly activated by TCF–Arm (Fang et al, 2006; Chang et al, 2008), these results indicate that additional sequence information must exist in these elements that influence CtBP's relationship with TCF and Arm. Our findings that CtBP is required for activation of a simple UAS-luc reporter by Gal4–Arm^* and Gal4–Pygo fusion proteins (Fang et al, 2006; Figure 4E and F), suggests that activation by CtBP might be the default state for the Wg pathway.

How can CtBP both promote and repress transcription of Wg targets? Our data demonstrate that the quaternary state of the CtBP protein determines its role as an activator and repressor. CtBP mutants that cannot homo-dimerize are unable to repress nkd-UpE1 expression (Figures 5C and 6D) or inhibit wing margin formation (Table I). However, co-expression of complimentary monomeric CtBP mutants that can hetero-dimerize restores CtBP repression activity in both these readouts (Figure 6D and Table I). This provides a compelling argument that self-association of CtBP is required for its ability to antagonize Wg transcriptional targets.

In contrast to targets where CtBP inhibits Wg signalling, monomeric CtBP can rescue the loss of Wg activation of the CG6234 WRE reporter in Kc cells depleted of endogenous CtBP (Figure 4C and D). In addition, CtBP monomers enhance an Arm-induced small eye phenotype (Figure 2C), can rescue a weak loss of Wg signalling defect in the fly wing (Figure 2G) and activate the Wg target Dll-lacZ (Figure 3; Supplementary Figure S1). Taken together, our data strongly support a model where CtBP monomers activate and CtBP dimers repress the Wg pathway in a gene-specific manner.

Normal or monomeric CtBP promote expression of UAS-luc, CG6234-luc and Dll-lacZ (Figures 3 and 4). In contrast, wild-type or hetero-dimeric CtBP represses the nkd-UpE1 reporter while monomeric CtBP cannot (Figures 5 and 6). The situation for CtBP's action on eye size in a GMR-Arm^* genetic background is more complicated, with WT CtBP suppressing the reduction in eye size (Figure 2B) while monomeric CtBP enhances the GMR-Arm^*-induced small eye phenotype (Figure 2C). The same is true for the wing margin, where monomeric CtBP can rescue the loss of wing margin caused by a reduction in Wg signalling (Figure 2D–G), while WT CtBP promotes wing notching (Table I). We suspect that unlike other Wg readouts, where either the activation or repressing function of CtBP is dominant, in the GMR-Arm^* eye and at the wing margin, both activities are prevalent, with the repressive function winning out in oligomeric CtBP expression and the activation function with monomeric CtBP.

Mechanism of CtBP action on Wg signalling

Why is dimerization required for repression by CtBP? CtBPs have been shown to bind to many transcriptional repressors and some interactions require the dimerization of CtBP (Turner and Crossley, 2001; Kumar et al, 2002; Balasubramanian et al, 2003; Chinnadurai, 2007; Kuppuswamy et al, 2008). Although homo-dimerization is dispensable for interaction of CtBP with some factors, it is clearly required for the function of CtBP as a potent co-repressor in complex with those factors (Kuppuswamy et al, 2008; Zhao et al, 2009). CtBPs have also been shown to associate with several chromatin-modifying enzymes, which have crucial roles in transcriptional repression (Shi et al, 2003; Kuppuswamy et al, 2008; Zhao et al, 2009). A recent report provides good evidence that in the case of repression of E-Cadherin by the repressor ZEB, human CtBP2 dimers act as adaptors between ZEB and histone deacetylase 2 (HDAC2) (Zhao et al, 2009). Binding of ZEB and HDAC2 to CtBP monomers was mutually exclusive, necessitating the need for CtBP dimerization to form the ZEB–CtBP–HDAC2 complex (Zhao et al, 2009). It is possible that this mechanism is also operating in the silencing of nkd expression in Kc cells, though further work is required to identify the binding partners of CtBP in this system.

In the case of activation, we have previously shown that CtBP functionally interacts with the N-terminal transactivation domain of Arm (Fang et al, 2006). This domain of Arm is bound by Lgs, which is in a complex with Pygo to promote transcriptional activation (Kramps et al, 2002; Stadeli and Basler, 2005; Li et al, 2007). Here, we extend our understanding of the mode of action of CtBP in activating Wg targets by demonstrating that CtBP substantially contributes to transcriptional activation by a Gal4–Pygo fusion protein (Figure 4F). Since Gal4–Pygo acts downstream of Arm and Lgs in activating transcription (Stadeli and Basler, 2005), our results indicate that CtBP acts downstream of Pygo.

The requirement of CtBP for Pygo activity is interesting in light of the genetic interaction data between pygo and monomeric CtBP in the developing wing. Overexpression of pygo causes notches in the wing margin, which is partially rescued by expression of monomeric CtBP (Figure 2D–G). Expression of pygo blocks Wg signalling in several contexts (Parker et al, 2002), presumably by disrupting the stoichiometry of a Lgs–Pygo–unknown factor(s) complex (i.e. shifting the equilibrium to Lgs–Pygo and Pygo–unknown factor hetero-dimers). The data in Figure 2D–G are consistent with a model where expression of monomeric CtBP shifts the equilibrium back to a trimeric, Lgs–Pygo–CtBP complex, which can promote activation of Wg targets in the wing margin.

The Lgs–Pygo complex is generally required for Wg signalling throughout fly development (Belenkaya et al, 2002; Parker et al, 2002; Thompson et al, 2002). In Kc cells, Pygo is required for activation of both CG6234 and nkd by Wg signalling (Parker and Cadigan, unpublished data). This suggests that there are other factors involved to explain why only a subset of Wg targets require CtBP for activation. In addition, Pygo has also been demonstrated to regulate some Wg targets in the absence of signalling, suggesting a possible link with CtBP in this context (de la Roche and Bienz, 2007; Mieszczanek et al, 2008).

Regulation of the oligomeric state of CtBP

Several studies have demonstrated that CtBP can self-associate (Kumar et al, 2002; Balasubramanian et al, 2003; Nardini et al, 2003; Shi et al, 2003; Thio et al, 2004; Kim et al, 2005; Mani-Telang et al, 2007; Kuppuswamy et al, 2008; Zhao et al, 2009). However, the presence of a monomeric pool has not been as extensively demonstrated (Kim et al, 2005; Zhao et al, 2009). To assess the distribution of monomers versus oligomers of CtBP, differentially tagged forms of CtBP were expressed. Even a 10-fold higher dose of CtBP-Flag could not completely co-immunoprecipitate CtBP-V5 (Figure 7A), indicating that a significant pool of CtBP-V5 is present as monomers or homo-dimers. Because similar results were obtained using CtBP^Basic and CtBP^Acidic-tagged forms (Figure 7B), which prevent homo-oligomerization, our data support a model where monomeric and homo-oligomeric pools of CtBP are present in Kc cells.

CtBPs are highly homologous to NAD⁺/NADH-dependent dehydrogenases and can bind NADH with high affinity (Fjeld et al, 2003). However, the role of co-factor based differential regulation of CtBP oligomerization is controversial. An increase in the NAD⁺/NADH levels stimulates dimerization of mammalian CtBPs (Kumar et al, 2002; Balasubramanian et al, 2003; Thio et al, 2004; Kim et al, 2005). Mutations in the NAD⁺/NADH-binding site of CtBPs abolish or reduce oligomerization (Kumar et al, 2002; Thio et al, 2004; Kuppuswamy et al, 2008; Nardini et al, 2009), although NAD⁺ binding is not always essential for dimerization of CtBP (Mani-Telang et al, 2007). In the case of the short isoform of fly CtBP, mutations in the NAD-binding cleft (G181V, G183V) make the protein highly unstable in Kc cells (Bhambhani and Cadigan, unpublished data). A more stable NAD cleft mutant (D204N) has nuclear localization defects and is unable to rescue CtBP mutant flies (Zhang and Arnosti, 2011). This precludes any functional studies to test if the NAD⁺/NADH ratio might affect its role in regulating Wg targets.

Does stimulation by Wg influence the oligomeric state of CtBP? In the absence of Wnt signalling, TCF acts with many other co-repressors to silence target gene expression. This repression is then counteracted by Arm/β-cat binding to TCF (Parker et al, 2007; Mosimann et al, 2009). Given the fact that CtBP dimers repress some Wg targets and CtBP monomers promote the Wg-dependent activation of some targets, it is tempting to speculate that Wg signalling causes a conversion of CtBP dimers to monomers. However, we have been unable to detect any difference in CtBP self-association with or without pathway activation in our co-IP assay (Supplementary Figure S6). Perhaps a more sensitive assay is required to detect changes in the oligomeric state upon Wg signalling.

An alternative to the Wg pathway influencing the oligomeric state of CtBP is a model where a pool of CtBP monomers and dimers exists in the cell in equilibrium (Figure 7). These pools might be differentially recruited to Wg targets by selective protein–protein interactions. In the case of CtBP monomers, this recruitment to WRE is predicted to require Arm. Although Wg signalling does not appear to influence the overall CtBP concentration in fly tissues (Fang et al, 2006), protein–protein interactions may cause changes in the monomer–dimer ratio on the WRE chromatin.

The role of CtBP oligomerization in other systems

The reagents and methodology described in this report can be applied to other systems where CtBP has important roles in regulating gene expression. For example, loss of CtBP1 and CtBP2 in the mouse results in loss of posterior structures in the embryo, a phenotype that has many similarities to Wnt3a mutants (Hildebrand and Soriano, 2002). This suggests that mammalian CtBPs also have a positive role in Wnt signalling. However, it is also possible that the phenotype is indirect, that is, CtBP represses a negative regulator of the Wnt pathway. Similar to fly CtBP, if murine CtBP monomers also have a positive role in regulating Wg targets, then a gene knockin of monomeric mCtBP1 or mCtBP2 should rescue the defect in posterior structures of CtBP knockouts.

In fly embryogenesis, loss of CtBP results in dramatic disruption of segmentation, due to defects in the striped pattern of the primary pair rule genes (Poortinga et al, 1998; Nibu et al, 1998a; Strunk et al, 2001). Many of these defects can be explained by the requirement of CtBP to bind to gap gene transcription factors (e.g. Kr) and promote repression (Nibu et al, 1998a; Keller et al, 2000; Nibu and Levine, 2001; Strunk et al, 2001; Struffi et al, 2004). However, there are aspects of the CtBP mutant phenotype (e.g. loss of pair rule stripes (Poortinga et al, 1998; Nibu et al, 1998a) and genetic interactions (Poortinga et al, 1998; Phippen et al, 2000) that suggest that CtBP may have a positive role in regulating transcription. Testing whether CtBP^Mono can rescue aspects of the CtBP segmentation phenotype may help determine whether CtBP has a direct role in activating transcription in regulatory hierarchies beyond the Wg pathway.

Drosophila cell culture

Kc167 cell culture and RNAi-mediated knockdown were performed as reported previously (Fang et al, 2006). Cells (10⁶/ml) were soaked in 10 μg dsRNA for 4 days, before seeding for transfections. Primers for dsRNA synthesis have been described elsewhere (Fang et al, 2006). Transient transfections were performed using Fugene 6 (Roche Applied Science) as per the manufacturer's instructions.

Plasmids and reporter assays

pAcCtBP^short with 2x Flag tags at the C-terminus (kindly provided by Dr D Arnosti) was used for all rescue assays. Site-directed mutagenesis of pAcCtBP^short (hereafter referred to as CtBP^WT) was used to introduce mutations in the dimerization interface to generate CtBP^Mono (C134Y, N138R, R141A, R142A), CtBP^Basic (E126R, E127R) and CtBP^Acidic (R171E, R173E). The C-terminal HA-tagged versions were generated by replacing the 2x Flag tags of pAcCtBP^short by 4x HA tags. The C-terminal V5-tagged versions were created by cloning the CtBP^WT and mutant cDNAs into the KpnI and NotI sites of pAC 5.1 V5-His (Invitrogen). pGL3nkd-UpE1, pAcArm*, pAcGal4DBD, pAcGal4Arm*, pUAS-luc and pActinlacZ constructs have been described elsewhere (Fang et al, 2006; Chang et al, 2008). pACGal4Pygo was constructing by cloning the dPygo ORF (815 amino acids) into the KpnI and XhoI sites of pACGal4DBD. pGL3CG6234 minimal WRE (CG6234 WRE), a 539-bp fragment, was generated using PCR-based subcloning of a 617-bp region (−3220 to −2603 relative to the CG6234 transcription start site) from the previously described pCG6234 (Fang et al, 2006). Deletion of the region (−2603 to −1465) and an internal 80 bp deletion (−2860 to −2781) led to ∼4.5-fold increase in the activation of the WRE by Arm^* in cell culture assays, and hence this reporter was used thereafter. For transgenic lines, cDNAs for the CtBP^WT and mutants with two C-terminal flag tags were subcloned into pUAST vector using the KpnI and XbaI sites.

CG6234 WRE and nkd-UpE1 reporter assays were performed by transiently transfecting 10–50 ng of the reporter and 150–500 ng CtBP expression plasmids in 2.5 × 10⁵ cells/well. The CG6234 WRE was activated using 100 μl of Wg-CM (derived from stable pTubWg S2 cells kindly provided by Dr R Nusse) or 50–100 ng of Arm^*. For assays with pUAS-luc, 10 ng of the reporter and 5 ng of Gal4Arm^* was used with 500 ng to 1 μg of the CtBP constructs. CtBP^Mono protein was less stable compared with CtBP^WT and to achieve equal expression levels, two times more of the CtBP^Mono plasmid was transfected compared with CtBP^WT in all assays. For all reporter assays, 5 ng of pAcLacZ was transfected for normalization and pAC5.1 (Invitrogen) or Gal4DBD to control for DNA amounts. Luciferase and LacZ assays were performed as described (Fang et al, 2006).

Drosophila genetics

Transgenic UAS-CtBP lines were generated using the injection facility at BestGene Inc. (Chino Hills, CA). w¹¹¹⁸, GMR-Gal4, Ptc-Gal4, Dpp-Gal4, En-Gal4 and Dll-lacZ were obtained from Bloomington Stock Center. CtBP transgenes were analysed for their effect on the small eye phenotype of P[GMR-Gal4] P[GMR-Arm^*] flies as described previously (Cadigan et al, 2002; Parker et al, 2002). Experiments with Ptc-Gal4, GMR-Gal4 and En-Gal4 were carried out at 25°C and Dpp-Gal4 at 27 or 29°C.

Immunoblots, immunostains and image quantification

For western blot analysis, anti-Flag (1:2500, Sigma), anti-V5 (1:5000, Invitrogen), anti-HA (1:1000, Roche) and anti-Tubulin (1:4000, Sigma) were used followed by HRP-conjugated anti-mouse or anti-rat IgG (Jackson Immunochemicals). Signal was detected using ECL kit (Amersham). Immunostaining of wing imaginal discs was performed as described previously (Fang et al, 2006) using anti-LacZ (1:1000, Abcam), anti-Wg (1:100) and anti-CtBP (1:1000). Alexa 488- and Cy3-conjugated secondary antibodies were obtained from Molecular Probes and Jackson Immunochemicals. Samples were examined using a Leica triple channel confocal microscope DM6000B-CS and processed using Adobe Photoshop 8.0.

For image quantification, total pixel intensity was determined using the Volocity Software 5.0 (Perkin-Elmer) after background subtraction. A region at the D/V boundary with the Dll-lacZ expression was selected in the anterior and posterior of the wing imaginal disc (Supplementary Figure S1). The mean pixel intensity of the posterior was normalized to the anterior.

Co-IP and ChIP

For co-IPs, 6–10 × 10⁶ Kc cells were seeded with 1 μg pAcCtBP/10⁶ cells for 3 days before harvesting. Cells were resuspended in lysis buffer (150 mM NaCl, 50 mM HEPES, pH 7.9, 1% CHAPS, 10% glycerol, 0.1 mM EDTA with complete mini-EDTA free protease inhibitor cocktail, (Roche) and sonicated thrice on ice in pulses of 6 s. An alternate lysis protocol was used for experiments in Figure 7. Cells were resuspended in lysis buffer (100 mM potassium phosphate (pH 7.8), 0.2% Triton X-100, 0.1 mM EDTA with complete mini-EDTA free protease inhibitor cocktail, (Roche) and incubated at 4°C for 1 h. Lysates were pre-cleared using Protein A/G sepharose beads. Total protein concentration was measured using the DC protein assay (Bio-Rad). Lysates corresponding to 3 mg total protein was used for each IP. In all, 15% of this lysate was saved as input. The remainder was incubated with 5 μg primary antibody for 2 h at 4°C followed by incubation with Protein A/G sepharose beads for 30 min at 4°C. The antibody–antigen complexes were washed four times with lysis buffer and eluted in 60 μl of Laemmli sample buffer for western blot analysis. Results shown are representative of at least two independent experiments.

ChIP analysis was performed as described previously (Fang et al, 2006). Briefly 3 × 10⁶ cells and 10 μl of anti-CtBP antisera were used for every pull down and precipitated DNA subject to quantitative RT–PCR. Data are expressed as a percent of the input DNA. Specific primer pairs for the UPE and ORF correspond to N#1 and N#0 in the nkd locus as reported elsewhere (Fang et al, 2006; Chang et al, 2008).