Plakoglobin and β-catenin differ in their ability to form a ternary complex with LEF/TCF factors and DNA.
To compare the ability of β-catenin and plakoglobin to form ternary complexes with LEF/TCF and DNA, we used the DNA mobility shift analysis (electrophoretic mobility shift assay [EMSA]) with a radiolabeled LEF-1 DNA binding site as the DNA probe. First, in vitro-transcribed and -translated LEF-1, β-catenin, and plakoglobin and deletion mutants of the two catenins (Fig. ) were prepared (Fig. A), and the interaction of LEF-1 with catenins and DNA was determined (Fig. B). Incubation of in vitro-translated LEF-1 with the DNA probe, in the absence of catenins, resulted in the formation of a single LEF-1–DNA band (Fig. B, lane 2). When LEF-1 was incubated with the DNA in the presence of β-catenin (at a fivefold molar excess of β-catenin over LEF-1), an additional slower-migrating band was formed (Fig. B, lane 3), corresponding to the β-catenin–LEF-1–DNA complex. In contrast to β-catenin, plakoglobin did not form a detectable complex with LEF-1 and DNA (Fig. B, lane 4). When β-catenin and plakoglobin were compared in their ability to form catenin–TCF-4–DNA ternary complexes, plakoglobin was again found unable to form a complex composed of plakoglobin–TCF-4–DNA (Fig. C, lane 3), while a β-catenin–TCF-4–DNA complex was efficiently formed under the same conditions (Fig. C, lane 2, compare to lanes 1 and 3). These results demonstrate that the observed difference between β-catenin and plakoglobin in the efficiency in forming a ternary complex with LEF/TCF factors and DNA was apparent with both LEF-1 and TCF-4 and was not unique to LEF-1.
FIG. 1 Schematic representation of catenin constructs used in this study. Catenin domains that have transactivation capacity are hatched. The numbers 1 to 13 represent armadillo (ARM) repeats in β-catenin and plakoglobin with a nonrepeat region (ins) (more ...)
FIG. 2 Identification of β-catenin and plakoglobin domains governing the formation of catenin–LEF-1–DNA complexes. (A) The catenin constructs and LEF/TCF factors were transcribed and translated in vitro in the presence of [35 (more ...)
Since the armadillo repeat domain of β-catenin was implied to contain the binding site for LEF/TCF (4
), we next compared the abilities of deletion mutants of β-catenin and plakoglobin comprised of the armadillo repeat domains of these proteins (Fig. , β-CAT ARM and PG ARM) to those of full-length catenins in forming ternary complexes by EMSA. The results shown in Fig. B demonstrate that there was a significant difference in catenin–LEF-1–DNA complex formation between these constructs: both β-CAT ARM and PG ARM were substantially more efficient in forming catenin–LEF-1–DNA complexes than were the respective full-length molecules (Fig. B, lanes 9 and 10, compare to lanes 3 and 4). To determine whether the N or the C termini of catenins were responsible for conferring the inefficient capacity to form ternary complexes by the full-length catenins, we prepared ΔN and ΔC deletion mutants of both β-catenin and plakoglobin (Fig. , β-CAT ΔC, β-CAT ΔN, PG ΔC, and PG ΔN) and tested their ability to form complexes with LEF-1 and DNA. Both ΔN and ΔC β-catenin mutants interacted with LEF-1 with an efficiency that was similar to that of the full-length protein but which was much lower than that of the armadillo repeat domain (Fig. B, lanes 5 and 7, compare to lanes 3 and 9). ΔN- and ΔC-plakoglobin constructs were also significantly less effective in their interaction with LEF-1 than was PG ARM (Fig. B, lanes 6 and 8, compare to lane 10). PG ΔC was slightly more efficient than full-length plakoglobin in forming such a complex (Fig. B, compare lane 4 to 8, which displays a smear). Thus, both the N- and C-terminal domains of β-catenin and plakoglobin could apparently act as negative regulators of catenin–LEF-1–DNA ternary complex formation. Removal of the last three armadillo repeats of both β-catenin and plakoglobin (Fig. , β-CAT 1-ins and PG 1-ins) completely abolished the interaction with LEF-1 and DNA (Fig. B, lanes 11 and 12), indicating that the integrity of the armadillo repeat domain is necessary for binding to LEF-1 (see below).
To determine the contribution of the catenin terminal domains and the armadillo repeats to the difference in ternary complex formation between the two catenins, chimeras of β-catenin and plakoglobin were also prepared (Fig. A). These constructs were translated in vitro (Fig. B) and tested for the ability to form ternary complexes with LEF-1 (Fig. C). When the terminal domains of plakoglobin were replaced by those of β-catenin, the resulting chimera (β-N/PG ARM/β-C), similarly to full-length plakoglobin (Fig. C, lane 4), was unable to form a detectable complex with LEF-1 and the DNA (Fig. C, lane 5), thus being significantly different from β-catenin (Fig. C, lane 3). This implies that the terminal domains of either β-catenin or plakoglobin exert a similar inhibitory effect on ternary complex formation by PG ARM.
FIG. 3 Formation of catenin–LEF–DNA complexes by a chimera of β-catenin and plakoglobin. (A) Schematic representation of β-catenin and plakoglobin chimeric molecules. The β-catenin domains are shown in black, and those (more ...)
To determine the inhibitory properties of the C-terminal domains, in the absence of the N terminus, we tested chimeric molecules consisting of the ARM repeats of plakoglobin and β-catenin linked to the C-terminal domain of the other molecule (Fig. A, PG ARM/β-C and β-CAT ARM/PG-C). Comparison of these chimeras to β-catenin ΔN and plakoglobin ΔN showed that the β-CAT ARM/PG-C chimeric molecule behaved similarly to β-catenin ΔN (Fig. C, lanes 6 and 8) and that PG ARM/β-C was similar to plakoglobin ΔN (Fig. C, lanes 7 and 9), in their efficiency in forming ternary complexes. The β-N/PG ARM chimeric molecule formed a smear (Fig. C, lane 10) indicative of weak complexing, similar to that observed with plakoglobin ΔC (Fig. C, lane 8). Taken together, these experiments demonstrate that the inhibitory properties of the N- and C-terminal domains of β-catenin and plakoglobin are similar, and the major difference between the catenins in ternary complex formation is apparently determined by the ARM domains of these molecules (Fig. B, compare lanes 9 and 10).
β-Catenin–LEF-1–DNA but not plakoglobin–LEF-1–DNA complexes are formed in cells transfected with plakoglobin.
We have previously demonstrated that the transfection of plakoglobin into 293-T cells results in LEF-1-dependent transcriptional activation (70
). Since plakoglobin was inefficient in the formation of a complex with LEF-1 and DNA in vitro (Fig. B and C and Fig. C), we next asked whether it could form such a complex in the nuclei of cells overexpressing plakoglobin. 293-T cells were transfected with HA-tagged LEF-1 either alone or together with VSV-tagged plakoglobin, or β-catenin (Fig. A), and EMSAs with nuclear extracts from the transfected cells were conducted (Fig. B). Nuclear extracts from cells transfected with a control vector formed bands (Fig. B, lane 1, N. S.) that were competed by excess unlabeled DNA comprised of the mutant LEF-1 binding site and were thus considered nonspecific (results not shown).
FIG. 4 β-Catenin–LEF-1–DNA complex formation in plakoglobin-transfected cells. (A) Schematic representation of N- and C-terminally VSV-tagged catenin constructs. (B) Catenin–LEF-1–DNA complex formation was analyzed with (more ...)
When nuclear extracts from cells transfected with HA-tagged LEF-1 were analyzed, an additional strong band was detected (Fig. B, lane 2), which corresponded to the LEF-1–DNA complex and displayed a lower mobility in the presence of anti-HA antibodies (results not shown). When extracts from cells transfected with LEF-1 and β-catenin, or plakoglobin, were analyzed, an additional band (β-CAT–LEF–DNA) whose position corresponded to the expected position of a catenin–LEF-1 complex was apparent (Fig. B, lanes 3 and 6). To identify the molecular composition of this band, nuclear extracts were incubated with antibodies to the VSV tag, or to β-catenin (Fig. B, lanes 4 and 7 and 5 and 8, respectively). As expected, in extracts from β-catenin-transfected cells this band was supershifted by anti-VSV and anti-β-catenin antibodies (Fig. B, lanes 7 and 8, compare to lane 6), confirming that it contained the β-catenin–LEF-1–DNA complex. In contrast, with extracts from cells transfected with VSV-plakoglobin, only the anti-β-catenin antibody induced a shift in this band (Fig. B, lane 5), while the anti-VSV antibody did not alter the migration of this band (Fig. B, lane 4). Since the level of transfected plakoglobin in the nuclear extracts was higher than that of transfected β-catenin (Fig. C), these results imply that plakoglobin did not form a detectable complex with LEF-1–DNA in the VSV-plakoglobin-transfected cells (Fig. B, lane 4) but that, rather, the endogenous β-catenin became engaged in a complex with LEF-1 and the DNA.
Removal of the N- and C-terminal domains of β-catenin and plakoglobin enhances catenin–LEF-1–DNA complex formation in transfected cells.
To analyze the ability of the terminal domains of β-catenin and plakoglobin to regulate catenin–LEF-1–DNA complex formation in cells, nuclear extracts from cells transfected with LEF-1 and with VSV-tagged armadillo repeat domains of β-catenin and plakoglobin (Fig. A) were subjected to DNA mobility shift analysis. In these experiments, the strong band corresponding to the LEF-1–DNA complex (Fig. A, lane 3) was almost completely supershifted in extracts containing the armadillo repeat domains of β-catenin (Fig. A, lane 6) and plakoglobin (Fig. A, lane 7) compared to full-length catenins (Fig. B, lanes 3 and 6). This demonstrates the very efficient formation of DNA-bound complexes comprised of LEF-1–DNA and the armadillo domains. To confirm that these complexes indeed contained β-CAT-ARM-VSV and PG-ARM-VSV, the anti-VSV antibody was added to the nuclear extracts. This resulted in the induction of a supershift in the migration of these bands (Fig. A, lanes 10 and 11, compare to lanes 6 and 7). The addition of antibodies to β-catenin to such lysates did not induce an efficient supershift (Fig. A, lanes 14, 15), indicating that the nuclear extracts from cells cotransfected with LEF-1 and the armadillo repeat domains of catenins did not contain significant amounts of endogenous β-catenin complexed with LEF-1. Since the polyclonal antibody to β-catenin that was used here did not react with the armadillo repeat domain of β-catenin (data not shown), we attributed the very weak supershift observed with this antibody (Fig. A, lanes 14 and 15) to a ternary complex formed by the endogenous β-catenin.
FIG. 5 The armadillo repeats of β-catenin and plakoglobin efficiently form ternary complexes with LEF-1 and DNA in transfected cells. (A) EMSA was conducted with nuclear extracts from 293-T cells transfected with control vector (lane 2) or with LEF-1 (more ...)
The VSV-tagged β-catenin used in this experiment (Fig. A, lane 4) contained an N-terminal VSV tag (Fig. A), in contrast to the C-terminal VSV tag used in the experiments described for Fig. B. This resulted in a different efficiency in the supershifts induced by the anti-VSV and anti-β-catenin antibodies in the two experiments (compare Fig. B, lanes 7 and 8, to Fig. A, lanes 8 and 12). It is noteworthy that in nuclear extracts from transfected cells, PG-ARM-VSV was an efficient partner for LEF-1–DNA, while full-length plakoglobin did not form a detectable complex with LEF-1–DNA (Fig. A, lanes 7 and 11, compare to lanes 5 and 9), although the level of plakoglobin in such extracts was higher than that of the PG ARM, as determined by Western blotting (results not shown). These results demonstrate that the increase in catenin–LEF-1–DNA complex formation observed after removal of the N- and C-terminal domains using in vitro-translated components (Fig. B) was also apparent with nuclear lysates from cells transfected with these constructs.
To examine whether the differences observed in ternary complex formation resulted from different affinities of the various catenin constructs for LEF-1, we determined the interaction of catenins and their deletion mutants with LEF-1 by coimmunoprecipitation (Fig. B and C). VSV-tagged catenins were cotransfected with HA-tagged LEF-1, and immunoprecipitation was performed with either monoclonal anti-HA antibody (Fig. B) or polyclonal anti-VSV antibodies (Fig. C). In cells expressing similar levels of catenins and LEF-1 (Fig. B), the efficiencies of coprecipitation of full-length β-catenin and plakoglobin with LEF-1, by an antibody recognizing the transfected LEF-1 (HA), were very similar (Fig. B, lower panel, lanes 1 and 2). Similarly, the coprecipitations of LEF-1 with both catenins by anti-VSV that recognized the catenins were also similar (Fig. C, lower panel, lanes 2 and 3). These results, together with previously published data (29
), strongly suggest that the differences observed between β-catenin and plakoglobin in ternary complex formation with LEF-1 and DNA apparently do not result from different affinities of the catenins for LEF-1 but most probably are due to the weaker capacity of the plakoglobin–LEF-1 complex to interact with DNA.
We have also compared the abilities of full-length catenins to those of their deletion mutants (by coimmunoprecipitation) in the interaction with LEF-1. The results shown in Fig. C (lower panel, lanes 4 and 5) demonstrate that both ARM repeats efficiently bind LEF-1 while PG 1-ins does not display a detectable binding to LEF-1 (Fig. C, lane 6), in agreement with its inability to shift the LEF-1–DNA complex in EMSA (Fig. B, lanes 11 and 12). The low expression level of the VSV-tagged β-CAT 1-ins precluded us from using it in this experiment (results not shown).
Differential ability of catenin deletion mutants to elevate and translocate endogenous β-catenin into the nucleus.
The formation of a complex between endogenous β-catenin and LEF-1 and DNA, using extracts from cells transfected with catenin mutants (β-CAT-ARM-VSV and PG-ARM-VSV), suggested that these mutants could elevate endogenous β-catenin, similar to observations made with cells transfected with full-length plakoglobin (70
). To examine this possibility, we determined, by immunofluorescence microscopy, the abilities of the catenin deletion mutants to elevate and consequently alter the localization of endogenous β-catenin.
293-T cells were transiently transfected with deletion mutants of catenins and doubly stained for the endogenous β-catenin and for the transfected protein (Fig. and ). In general, the various deletion mutants displayed similar subcellular localizations: they were mainly localized in the nuclei of the transfected cells (Fig. C and G and 7A and E) and only sometimes also localized at cell-cell junctions (arrows in Fig. and ). In addition, PG 1-ins staining was mainly diffuse, while the other constructs were mostly confined to the nucleus.
FIG. 6 The effect of plakoglobin mutants on the localization of endogenous β-catenin. 293-T cells were transiently transfected with the HA-tagged plakoglobin constructs described in Fig. and doubly stained for the HA tag (A, C, E, and (more ...)
FIG. 7 Transfection of deletion mutants or membrane-tethered forms of β-catenin increases the level and translocates endogenous β-catenin to the nucleus. 293-T cells were transiently transfected with plasmids encoding the armadillo repeat of (more ...)
Double staining for the endogenous β-catenin showed that transfection with both β-CAT ARM (Fig. A) and PG ARM (Fig. C) resulted in nuclear translocation of the endogenous β-catenin (Fig. B and D, respectively), similar to the effect of full-length plakoglobin (Fig. A and B). This result is attributed to the compromised degradation of β-catenin in these cells (43
). The shorter truncation mutants, PG 1-ins and β-CAT 1-ins (Fig. ), differed in their effect on endogenous β-catenin (Fig. H and F), as previously described (70
). While transfection with β-CAT 1-ins (Fig. E) resulted in nuclear localization of the endogenous β-catenin (Fig. F), transfection with PG 1-ins (Fig. G) did not display a detectable effect on the endogenous β-catenin (Fig. H). This implies that the regions within the armadillo repeat domains that determine the binding of catenins to components of the degradation machinery are different in β-catenin and in plakoglobin.
To examine whether nuclear localization of the armadillo repeat domains of both catenins was necessary for nuclear translocation of the endogenous β-catenin, we also prepared fusion constructs of these molecules to the transmembrane domain of connexin (Fig. ). Such membrane-tethered armadillo repeats of β-catenin and plakoglobin were excluded from the nucleus and localized to cytoplasmic vesicles and to cell borders (arrowheads in Fig. E and C). The ability to affect the level and localization of endogenous β-catenin, however, was preserved in such constructs, and cells overexpressing these cytoplasm-anchored mutants displayed nuclear staining of the endogenous β-catenin (Fig. F and D).
We have also determined the fluorescence intensity of endogenous β-catenin in the nuclei and cytoplasm of cells transfected with plakoglobin and with various deletion mutants of catenins (Fig. G) and found that, except for PG 1-ins, which did not affect the level of endogenous β-catenin, transfection with the other catenin mutants moderately elevated the cytoplasmic pool of β-catenin (between 1.5- and 2.5-fold) and induced a two- to fourfold increase in the nuclear level of β-catenin (Fig. G). The connexin-anchored catenin mutants were the most efficient in their effect on endogenous β-catenin, most probably owing to their exclusive cytoplasmic localization. Interestingly, the level of junctional β-catenin was not significantly affected by transfection with the catenin constructs (Fig. and and data not shown).
Taken together, these results suggest that the β-catenin and plakoglobin mutants examined (except PG 1-ins), were capable of increasing the level and induced the translocation into the nucleus of the endogenous β-catenin, and their cytoplasmic sequestration did not affect this ability.
Transcriptional capacities of deletion mutants of β-catenin and plakoglobin.
The ability of the various catenin deletion mutants to increase the level and translocate endogenous β-catenin into the nucleus did not always correlate with their ability to form ternary complexes with LEF-1–DNA (Fig. B, compare to Fig. and ). For example, while both β-CAT ARM and β-CAT 1-ins could translocate the endogenous β-catenin into the nucleus (Fig. B and F), only β-CAT ARM, not β-CAT 1-ins, formed a complex with LEF-1 and DNA (Fig. B, lanes 9 and 11). We have therefore analyzed the transactivation capacity of these constructs. We expected that constructs lacking the transactivation domain, but which efficiently formed a complex with LEF-1 and the DNA, would act as dominant-negative inhibitors of transactivation by β-catenin, irrespective of their influence on endogenous β-catenin. We also expected that the transactivation capacity of catenin deletion mutants that do not form complexes with LEF-1 and DNA (β-CAT 1-ins and PG 1-ins) would depend on their capacity to elevate endogenous β-catenin (70
To test these predictions, 293-T cells were transfected with the various constructs, together with a reporter plasmid that detected LEF-1-responsive transcription, and the activity of this reporter was determined (Fig. A). As expected, neither β-CAT ARM nor PG ARM and PG 1-ins induced significant transcriptional activation of the reporter, while transfection of β-CAT 1-ins led to a 20-fold activation of the reporter (Fig. A). The transactivation potential of β-CAT 1-ins was similar to that of PG (Fig. A) (70
), in line with their similar capacities to elevate the level of endogenous β-catenin (Fig. and ). PG ARM and PG 1-ins were both weak in transactivation (Fig. A), while their abilities to form complexes with LEF-1–DNA were very different (Fig. B, lanes 10 and 12).
FIG. 8 Transcriptional properties of various β-catenin and plakoglobin constructs. 293-T cells were transfected with different catenin constructs together with a LEF-1-responsive reporter and the LacZ plasmid that served as control for transfection efficiency (more ...)
To examine whether an efficient complex formation by these mutants with LEF-1–DNA is reflected in a dominant-negative effect on β-catenin-directed transactivation, β-catenin was transfected together with the expression plasmids coding for β-CAT ARM, PG ARM, β-CAT 1-ins, and PG 1-ins (Fig. B). In agreement with the in vitro binding studies, both β-CAT ARM and PG ARM inhibited β-catenin-driven transactivation, while β-CAT 1-ins and PG 1-ins had no effect (Fig. B; see Fig. C for the proposed mechanism).
FIG. 9 Schematic representation of the proposed effect of plakoglobin and catenin mutants on endogenous β-catenin and LEF/TCF-dependent transcription. (A) When β-catenin levels are elevated, β-catenin activates transcription directly (more ...)
To demonstrate that this inhibition of transcription requires nuclear localization of the β-CAT ARM and PG ARM constructs, we have also used fusion proteins of these catenin mutants with the transmembrane domain of connexin (Fig. ). With these membrane-tethered forms of β-CAT ARM and PG ARM (CNX-PG ARM and CNX-β-CAT ARM), there was no inhibition of transactivation, but rather a significant induction of transcription (Fig. B). This was expected, as these fusion proteins were localized in the cytoplasm (Fig. C and E), elevated endogenous β-catenin levels, and efficiently activated transcription (Fig. A), most probably by the endogenous β-catenin (Fig. D).
Taken together, the transactivation studies demonstrated that the differences observed in vitro in LEF-1–DNA complex formation with the various β-catenin and plakoglobin mutants were also reflected in their different transactivation capacities in transfected cells.