Intracellular pH controls Wnt signaling downstream of glycolysis in the vertebrate embryo

Formation of the body of vertebrate embryos proceeds sequentially by posterior addition of tissues. While this process depends on aerobic glycolysis acting upstream of Wnt signaling in tail bud cells, the molecular details of this regulation are unknown. Here we used chicken embryos and human tail bud-like cells differentiated in vitro from iPS cells to show that glycolysis acts by increasing the intracellular pH (pHi) of tail bud cells. This promotes β-catenin acetylation leading to Wnt signaling activation and the choice of a mesodermal fate at the expense of the neural fate in tail bud precursors. Our data suggest that by increasing the pHi of tail bud cells, aerobic glycolysis creates a favorable chemical environment for non-enzymatic acetylation of β-catenin, ultimately triggering Wnt signaling.


Main text
Cells of the tail bud and posterior Presomitic Mesoderm (PSM) exhibit a high level of aerobic glycolysis which is reminiscent of the metabolic status of cancer cells experiencing Warburg effect (1). We previously reported that 2-day chicken embryos exhibit a posterior to anterior gradient of extracellular pH (pHe), with the lowest pH found in the tail bud where glycolysis is most active (2). Treatment of such embryos with the glycolysis inhibitor 2 Deoxy-D-glucose (2DG) or culture in glucose-free medium results in a uniform higher pHe consistent with a role for glycolysis in the acidification of the extracellular environment at the posterior end of the embryo (2). In normal differentiated adult cells, the intracellular pH (pHi) is usually around 7.2 whereas the pHe is generally around 7.4. In most cancer cells, the regulation of pHe and pHi is significantly different from normal cells (3): whereas pHe is lower (6.7-7.1) than in normal adult cells due to lactic acid excretion, pHi is higher (>7.4) compared to normal cells (3). To examine whether PSM and tail bud cells also exhibit an inverted pHe-pHi gradient, we measured the pHi in these tissues in chicken embryos. We used the ratiometric pH sensor pHluorin, a modified pHsensitive Green Fluorescent Protein (GFP) which has been previously used to analyze pHi in vivo (4,5). To test whether pHluorin can accurately report on pHi differences in the chicken embryo, we electroporated a pHluorin construct in the anterior primitive streak region (which contains the PSM precursors) of stage 4HH chicken embryos (6). One day after electroporation, embryos were incubated in buffers of different pH (pH5.5, pH6.5, pH7.5) with the protonophores Nigericin and Valinomycin. This allows the cells' pHi to equilibrate to the buffer pH (5). We next measured the average emission signal ratio at 488/405 nm in electroporated cells located in the tail bud and the PSM using confocal imaging and single cell 3D segmentation. As expected, we observed that the ratio decreased as the buffer pH increased (fig S1), indicating that the pHluorin reporter electroporated in vivo can report on intra-cytoplasmic pH changes in embryos. We next measured the 488/405 nm ratio in electroporated embryos in the absence of the protonophores. We observed a posterior to anterior gradient of pHluorin 488/405 nm signal ratio in the majority of wild type embryos (n=11/13) (Fig. 1A-B). With some local variability, the tailbud cells show a higher fraction of 405 nm excitable pHluorin whereas anterior PSM cells show a higher fraction of 488 nm excitable pHluorin (Fig. 1A-B). Thus, anterior PSM cells have higher intra-cytoplasmic acidity (lower pHi) compared to posterior cells.
We next explored whether glycolysis plays a role in the control of pHi in PSM cells. Treating embryos electroporated with pHluorin in the PSM with 2DG ex ovo increased the overall 488/405 nm ratio of the tail bud and posterior PSM (Fig. 1C), suggesting a global increase of intracellular acidity. Importantly, electroporated embryos grown on 2DG plates show no anterior-posterior gradient of 488/405 nm ratios compared to the embryos grown on control plates (Fig. 1D). We observed that culturing 2-day old chicken embryos in agar plates containing only PBS and 8.3 mM glucose at pH7.2 is sufficient to sustain normal development (2, 7) (Fig. 1E). Decreasing the glucose concentration to 0.83 mM and 0 mM in the minimal medium resulted in a dose-dependent slowing down of axis elongation similar to that seen with 2DG treatment (2) (Fig. 1E, fig. S2). We next cultured embryos in which pHluorin was electroporated in the PSM precursors in these low glucose conditions and measured the 488/405 nm signal ratio in the PSM. Decreasing the glucose concentration led to a dose-dependent increase of the 488/405 nm ratio, suggesting a decrease in pHi as observed when inhibiting glycolysis with 2DG (Fig. 1F). No significant pHi gradient could be evidenced in embryos cultured in low glucose conditions (Fig. 1G). Thus, decreasing glycolytic activity in cultured embryos leads to an increase of intracytoplasmic acidity in PSM cells and abolishes the anteriorposterior gradient of pHi. These data therefore support the idea that the observed spatial pH differences come from differential glycolytic activity along the PSM.
We next investigated the role of pHi in the control of PSM cell differentiation in vivo. Embryos electroporated with pHluorin were cultured ex ovo in minimal medium buffered at different pH with NaHCO3 for 3 hours and we measured the 488/405 nm signal ratio in the PSM ( Fig. 2A). The signal ratio in the posterior PSM decreased when the pH of the medium increased indicating that exposing embryos to a lower pHe leads to a corresponding acidification of the pHi in the PSM (Fig. 2B). When embryos were cultured at pH6.0, axis elongation was significantly slowed down (Fig. 2C, fig. S2). Culturing the embryos at pH5.3 led to a stronger phenotype with axis elongation rapidly stalling (Fig. 2C, fig. S2). These phenotypes are similar to those observed when inhibiting glycolysis with 2DG (2) or with low glucose concentrations (Fig. 1E). Remarkably, embryos exposed to a low pH buffer switched back to control medium after elongation arrest resumed normal development ( fig. S2). Thus, the developmental arrest observed after lowering the pH is not due to irreversible damage to the embryo. Together, these experiments suggest that appropriate regulation of pHi is essential for normal elongation of the embryonic axis.
In the tail bud, glycolysis is required for Wnt activation, which is necessary for axis elongation and for paraxial mesoderm production from the bi-potential neuro-mesodermal precursor (NMP) population (2,8,9). Blocking glycolysis with 2DG in vivo phenocopies Wnt signaling loss of function (2). This results in most NMPs to differentiate to a neural SOX2/SAX1-positive fate, ultimately leading to elongation arrest (2). Our data indicate that glycolysis is necessary to maintain a high pHi in the PSM/tail bud region and to support axis elongation, thus raising the possibility that glycolysis regulates Wnt signaling via the control of pHi. To test this hypothesis, we examined how altering tail bud pHi impacts the expression of the canonical Wnt target AXIN2 using in situ hybridization and qPCR ( Fig. 2D-E). Strikingly, culturing embryos in a low pH buffer decreases AXIN2 expression while culture at high pH results in stronger expression ( Fig. 2D-E). A similar behavior is observed for MSGN1, another Wnt/β-catenin target specifically expressed in the posterior PSM ( Fig. 2D-E) (10). In contrast, the neural genes SOX2 or SAX1 showed an opposite behavior, being upregulated at low pH ( Fig. 2D-E). Together, these results suggest that higher pHi favors canonical Wnt signaling, promoting the paraxial mesoderm fate from NMPs, while lower pHi inhibits Wnt, promoting the neural fate. Thus, these data argue that glycolysis could play a role in controlling Wnt signaling in tail bud cells by maintaining a high pHi.
We recently reported a protocol to differentiate human pluripotent cells to the paraxial mesoderm fate in vitro (11,12). Treatment of human iPS cells with the GSK3β inhibitor CHIR99021 (Chir) in combination with the BMP inhibitor LDN193189 (LDN) (CL medium), induces the majority of cells to differentiate to a SOX2-BRACHYURY positive NMP-like fate after day 1 (13). After day 2, more than 95% of the cells differentiate to a MSGN1-positive posterior PSM fate (11)(12)(13). In the following days, cells continue to differentiate, expressing the anterior PSM/dermomyotome marker PAX3 around day 6, and they can be subsequently differentiated to the myogenic lineage (11,12). We first examined whether the human iPS cells differentiating to the paraxial mesoderm lineage recapitulate the metabolic status of the tail bud and PSM cell populations observed in vivo in chicken and mouse embryos (2,14). We detected a decrease of lactate dehydrogenase (LDHB) and of lactate production after day 2 in culture ( Fig. 3A). This suggests that differentiating PSM cells downregulate glycolysis as described in the developing mouse and chicken embryos (2,14). In the mouse PSM, nuclear β -catenin is only found in the posterior region which exhibits high glycolytic activity (15). In the differentiating iPS cultures examined by immunohistochemistry, nuclear β -catenin localization (which is generally associated to Wnt activation (16)) peaked at day 2 ( Fig. 3B-C) (13). This correlated with the expression of the PSM Wnt targets AXIN2, TBX6 and MSGN1 measured by qPCR, which peaked around day 2-3 of differentiation ( Fig. 3D-F) (17)(18)(19). Thus, PSM cells exhibit an increased Wnt response during the early phase of PSM maturation in vitro, as observed in the posterior PSM in vivo (20). We next examined the pHi of iPS cells differentiating to the PSM fate in vitro using the pH sensitive ratiometric dye BCECF (5). We observed a higher pHi (>7.5) at day 2 progressively decreasing in sync with Wnt signaling after day 3 (Fig. 3G). Thus, the decrease of glycolytic and Wnt activity parallels a progressive decrease of pHi in human PSM cells differentiating in vitro. These data show that the metabolic status of mouse and chicken PSM cells is largely recapitulated in human iPS-derived PSM cells differentiated in vitro.
We observed that culturing iPS cells in CL medium at different pH can predictably change the pHi of cells (Fig. 3H). By qPCR analysis, we showed that decreasing the pHi by culturing cells in acidic conditions at a stage equivalent to the NMP stage (day 1) increased the expression of the neural marker SOX1 and decreased the expression of AXIN2 as was observed in vivo (Fig. 3I). Compared to acidic conditions, cells exposed to basic conditions (pH7.5 to 8) showed significantly decreased expression of SOX1 and increased expression of AXIN2 (Fig. 3I). Therefore, these data suggest that higher pHi promotes the paraxial mesoderm fate from tail budlike cells differentiating in vitro whereas lower pHi favors the neural fate in these cells.
We also noted that reducing glycolytic activity by treating differentiating iPS cells in vitro with 2DG decreases the pHi of cells (Fig. 3J). This is accompanied by an increase in SOX1 expression (Fig. 3K). This effect could be significantly rescued by incubating cells cultured in 2DGcontaining medium at a higher pH (Fig. 3K). In these conditions, SOX1 expression was decreased while AXIN2 was increased (Fig. 3K). A similar rescue was observed when cells were cultured in glucose free-medium (Fig. 3L). Thus, our in vivo and in vitro data suggest that pHi acts as a second messenger regulating Wnt signaling downstream of glycolysis in tail bud-like cells.
The dynamic regulation of Wnt signaling in differentiating iPS cells is somewhat unexpected as it occurs despite the constant presence of the Wnt activator Chir in the culture medium. Chir inhibits the kinase GSK3-β which phosphorylates β -catenin and targets it for degradation (16,21). This results in β -catenin stabilization allowing it to bind members of the Lef/TCF family which are required for shuttling β -catenin to the nucleus and for transcriptional activation of the Wnt targets (16). In whole extracts from cultures of human iPS-derived PSM cells, we observed a stable expression of β -catenin and its non-phosphorylated (active) form between day 1 to day 4, consistent with the expected effect of GSK3-β inhibition by Chir (Fig. 4A). Thus, the downregulation of nuclear β -catenin and Wnt targets observed in spite of Chir-treatment suggests that events downstream of β -catenin stabilization in the cytoplasm are dynamically regulated in PSM cells differentiating in vitro. One such candidate is K49 β -catenin acetylation, which acts as a switch controlling mesodermal vs neural gene activation in mouse embryonic stem cells (22). To test whether K49 β -catenin acetylation is associated with the peak of Wnt activation in vitro, we used an antibody which specifically detects K49 acetyl β -catenin. We observed a dynamic expression of K49 acetyl β -catenin peaking at day 1-3 in vitro (Fig. 4A). We analyzed the cytoplasmic and nuclear fractions of differentiating iPS cultures by western blot and observed a peak of nuclear K49 acetyl β -catenin around day 2 coincident with the peak of Wnt activation (Fig. 4B). Switching differentiating iPS cells to acidic conditions for 3 hours at day 2 led to a decrease of K49 acetylated β -catenin whereas exposure to alkaline conditions increased acetylated β -catenin levels compared to control (Fig. 4C). In human iPS cells differentiating to PSM, K49 acetylated β -catenin expression decreased in the absence of glucose (Fig. 4D). In all these different conditions, levels of non-phosphorylated and total β -catenin remained stable ( Fig.  4C-D). Stimulating global acetylation by treating differentiating cells with acetate, led to an increase of K49 β -catenin acetylation while global levels of β -catenin and its non-phosphorylated form remained stable (Fig. 4E). In addition, acetate treatment led to an upregulation of MSGN1 and a decrease in SOX1 expression, consistent with the PSM fate choice triggered by Wnt signaling activation (Fig. 4F). Thus, our data suggest that K49 β -catenin acetylation plays a role in the regulation of Wnt signaling during PSM differentiation in vitro.
Using an anti-acetyl-lysine antibody, we found that β -catenin is also acetylated in the PSM of chicken embryos (Fig. 4G). Exposure of 2-day chicken embryos to low pH (pH5.3 or 6.0) led to decreased acetylated β -catenin levels in the PSM while exposure to high pH (pH7.6) does not alter acetylated β -catenin levels compared to control embryos (Fig. 4G). Similar variations in acetylation levels were also observed with the antibody specific for K49 acetylation of β -catenin (Fig. 4G). In contrast, levels of active non-phosphorylated and of total β -catenin were stable when embryos were exposed to buffers of different pH (Fig. 4G, fig. S3). This indicates that acetylation of β -catenin is specifically affected by pH changes as observed in vitro. Reduction of acetylated and K49 acetyl but not total or non-phosphorylated β -catenin levels was also observed when reducing glucose concentration in embryos cultured in minimal medium (Fig. 4H). Conversely, treatment of 2-day chicken embryos with Sodium Acetate led to an upregulation of K49 β -catenin acetylation but not of total or non-phosphorylated β -catenin levels (Fig. 4I). This also resulted in an upregulation of AXIN2 and MSGN1 (Fig. 4J-K). Together these data show that β -catenin acetylation levels in the chicken embryo PSM can be modulated by pHi, glucose or acetate concentration independently of total and non-phosphorylated β -catenin levels. Together, our results suggest that in the tail bud, the regulation of pHi downstream of glycolysis can tune β -catenin acetylation to control Wnt activation.
How can upregulation of the pHi lead to an increase of β -catenin acetylation? The pHi can control the protonation of specific histidines in proteins acting as pH sensors, leading to changes in the protein properties (3). In some epithelial cells, the pHi level can regulate β -catenin stability by controlling its binding to β -TrCP and hence its degradation (23). While acetylation of protein substrates such as histones is largely regulated by HATs and HDACs enzymes, non-enzymatic acetylation of proteins has also been demonstrated in many different cell types (24). This chemical addition of acetyl-residues to cellular proteins can be regulated by the pHi of cells (25). This raises the interesting possibility that the high pHi observed in PSM cells downstream of glycolysis might provide favorable chemical conditions for non-enzymatic β -catenin acetylation. To test this possibility, we performed an in vitro reaction, incubating recombinant β -catenin protein with acetyl-CoA sodium salt in PBS at pH5.0, pH6.3, pH7.2, and pH7.6 for 3 hours at 37C. Using an anti-acetyl lysine antibody, a clear dose-dependent increase of β -catenin acetylation was detected when the pH was raised (Fig. 4L). Strikingly, even small pH increments within a physiological range (pH6.5 to pH 7.5) led to significant dose-dependent increase of β catenin acetylation (Fig. 4M-N). Together our observations suggest a model wherein the peak of glycolysis in the tail bud/PSM imposes a high pHi to these cells thus favoring non-enzymatic β catenin acetylation downstream of Wnt signaling. This could ultimately lead to activation of mesodermal transcriptional Wnt targets and specification of the paraxial mesoderm.
Our data is also consistent with posterior PSM and tail bud cells exhibiting Warburg-like metabolism with high aerobic glycolysis and an inverted pHe/pHi gradient as compared to more differentiated cells. Decreasing the pHi in Wnt-addicted tumor cells was recently shown to inhibit Wnt signaling, as reported here for differentiating tail bud cells (26). Thus our findings further emphasize the tight similarity between the developing tail bud cell and cancer cells that exhibit high Warburg metabolism resulting in high pHi and low pHe (27), supporting the notion that some tumor cells might reactivate a developmental metabolic program. (E) Increase in axis length (elongation) measured over time using time lapse microscopy (mean ±SD) in the chemically defined medium at pH7.2 with 8.3mM glucose (circles, n=6), 0.83mM glucose (squares, n=3) and 0 mM glucose (triangles, n=6). Normal development is observed in the chemically defined medium although axis elongation is slower compared to normal Embryo Culture medium (EC) (85 μm/h vs 170 μm/h in normal EC culture). Statistical significance was assessed with two way Anova followed by Tukey's multiple comparison test. ***p<0.001.    (G) Analysis of the pHi using the BCECF dye in human iPS cells differentiating to the PSM fate in vitro in CL medium. Error bars represent SD (n=3). Statistical significance was assessed with one-way ANOVA followed by Tukey's multiple comparisons test. **p < 0.01.
(H) Analysis of the pHi using the BCECF dye in day 2 human iPS cells differentiated to the PSM fate in vitro in CL medium and cultured for 3 h in CL culture medium at different pH (n=4 for each condition). Error bars represent SD. Statistical significance was assessed with two-way ANOVA followed by Tukey's multiple comparisons test: *p < 0.05; ***p< 0.001; ****p<0.0001; n.s. p> 0.5 (I) Comparison of SOX1 and AXIN2 mRNA expression in day 2 human iPS cells differentiated to a PSM fate in vitro in CL medium and cultured for 3 h in CL medium at different pH (Control: pH7.0-7.2), Acid: pH6.3-6.5; Alkali: pH7.5-7.8) (n>=5 for each condition). Error bars represent SD. Statistical significance was assessed with two-way ANOVA followed by Tukey's multiple comparisons test: *p < 0.05; **p< 0.01.
(J) Analysis of the pHi using the BCECF dye in human iPS cells cultured for one day in CL medium followed by 24 h in CL medium with 2DG (n=4 for each condition). Error bars represent SD (n=4). Statistical significance was assessed with unpaired two-tailed test: *p < 0.05.
(K-L) Comparison of SOX1 and AXIN2 mRNA expression in human iPS cells cultured for one day in CL medium followed by 24 h in CL culture medium with 2DG (K) or in glucose free CL medium (L) at normal pH or in alkaline conditions (pH7.5-7.8) (n=3 experiments for each condition). Error bars represent SD. Statistical significance was assessed with two-way ANOVA followed by Tukey's multiple comparisons test: *p < 0.05; **p< 0.01.

Time-lapse microscopy and axis elongation measurements
Stage 9HH chicken embryos were cultured ventral side up on a microscope stage using a custom built time-lapse station (28). We used a computer controlled, wide-field (10× objective) epifluorescent microscope (Leica DMR) workstation, equipped with a motorized stage and cooled digital camera (QImaging Retiga 1300i), to acquire 12-bit grayscale intensity images (492 × 652 pixels). For each embryo, several images corresponding to different focal planes and different fields were captured at each single time-point (frame). The acquisition rate used was 10 frames per hour (6 min between frames). To quantify axis elongation length, the last formed somite at the beginning of the time-lapse experiment was taken as a reference point, and the position of the Hensen's Node with respect to this somite was tracked as a function of time using the manual tracking plug-in in Image J (29).

Plasmid preparation and electroporation for chicken embryo
The ratiometric pH sensor pHluorin (4) was used to generate the expression vector pCAGG-pHluorin-IRES2-Td-Tomato. Full length pHluorin sequence was sub-cloned in pENTR-1A vector (Invitrogen), and inserted into a pCAGGS-IRES2-tdt-RFA destination vector using the Gateway system (Invitrogen). Chicken embryos ranging from stage 6HH to stage 7HH were prepared for EC culture. A DNA solution (1.0-5.0 μ g/μl) was microinjected in the space between the vitelline membrane and the epiblast at the anterior primitive streak level which contains the precursors of the paraxial mesoderm. In vitro electroporations were carried out with five successive square pulses of 8V for 50ms, keeping 4mm distance between anode and cathode using Petri dish type electrodes (CUY701P2, Nepa Gene, Japan) and a CUY21 electroporator (Nepa Gene, Japan).

Intra-cytoplasmic pH measurement for chicken embryo
In order to measure the intra-cytoplasmic pH (pHi) in PSM and tail bud cells, stage 4-5 HH chicken embryos were electroporated with the pCAGGS-pHluorin-IRES2-Td-Tomato construct which contains the ratiometric pH biosensor pHluorin (4) and cultured until stage 12 HH at 38 °C. Embryos strongly expressing the constructs were subsequently selected based on Td-Tomato expression in the PSM and tail bud. We first tested whether pHluorin can accurately report on pHi differences when electroporated in the chicken embryo. Embryos were electroporated at the anterior primitive streak level with the pHluorin construct to target tail bud and PSM cells as described previously (34) and they were reincubated until they reached stage 12 HH. Embryos were then incubated in different buffers (pH5.5, pH6.5, pH7.5) from the pH calibration buffer kit (molecular probe) with 10 μM of the protonophores Nigericin and Valinomycin at 37 °C for 30 minutes. Exposure to the protonophores allows the cells' pHi to equilibrate to the buffer pH. Then embryos were mounted on MatTek glass-bottom dishes soaked in pH calibration buffer (pH5.5, pH6.5, pH7.5). Images were captured using a laser scanning confocal microscope (TCS SP5; Leica or LSM780; Zeiss) at 37 °C in humidified atmosphere. The protonophore treatment completely abolished the gradient of pHluorin 488/405 signal ratio and the average emission signal ratio under 488/405nm excitation decreased as pH increased ( fig. S3), indicating that the pHluorin reporter electroporated in vivo can report on intracytoplasmic pH changes. However, because it is impossible to perform the different steps required for calibration in the same embryo, different electroporated embryos were used for each pH value and for experimental measurements. Therefore, the absolute value of the pH could not be defined (Fig. 1A). Thus, for all in vivo measurements, we only compared the 405 to 488 nm fluorescence ratios.

Culture medium preparation
To examine the effect of medium pH, cells were incubated in media buffered at different pH for 3 hours at differentiation day2. To adjust the medium pH, HCl and Sodium bicarbonate (Sigma #S5761) were used and the pH of the medium was measured using a pH meter (Mettler toledo) or pH indicator papers (pH6.0-8.1) (GE Healthcare Life Sciences #2629-990). Prior to experiments, the medium pH was equilibrated by incubation in 5% CO2 at 37℃ for 24 h.
To examine the effect of glucose-free condition, cells were incubated in glucose (-) medium for 6h at differentiation day2 or for 24h at differentiation day1. 5mM 2DG (Sigma #D8375) was used in 5mM D-(+)-Glucose (Sigma # G7021) containing DMEM medium (#D9807-02) to inhibit glycolysis. Culture in CL medium with 2 or 10 mM Sodium acetate (Sigma #S2889) for 24 hour at differentiation day1 was used to increase the acetylation level.

Lactate assay
To detect lactate production by the differentiated iPS cells, we used a lactate assay kit (Bio Vision #K607-100) according to the manufacturer's instructions. Measurement of O.D. at 560nm was performed using GloMax-Multi Detection System (Promega). Values were normalized by total protein amount in each well.

Measurement of intracellular pH
The BCECF pH sensitive dye (ThermoFisher Scientific #B1150) was used to analyze the intracellular pH in human iPS cells according to the manufacturer's instructions. After washing the cells with HBSS (gibco #14025-092), 1μM AM ester solution (diluted with HBSS) was added to the culture and incubated for 20 minutes at 37℃. Then, cells were washed twice with fresh culture medium and incubated for 10 min at 37℃. After replacing the medium with HBSS, the 488/405 nm ratio was measured using confocal microscopy (LSM780, Zeiss) or GloMax-Multi Detection System (Promega). To determine the absolute pH value, we incubated different wells from the same experiment in the calibration buffer (3 different pH PBS adjusted by HCl or NaOH) with 10μM of the protonophores Nigericin and Valinomycin at 37℃ for 15minutes. Then, the 488/405 nm ratios were measured using confocal microscope or Glo-Max multidetection System and used to generate a calibration curve from which wells from the same experiments could be compared.

Quantification of nuclear localization
Nuclear localization of β -catenin was measured in cultures immunostained with the anti-betacatenin antibody using the Fiji software. First, we splitted the channels and smoothened the Hoechst staining images using the "Gaussian Blur" filter. Then, we binarized the image, cut the border of adjacent cells using "watershed" and eliminated the β -catenin expression in outermost of nucleus area using the "erode" filter. We next manually drew the contours of the nucleus and used the "analyze particles" tool to measure the signal intensity of β -catenin in each nucleus.

Nuclear fraction
Nuclear and Cytoplasmic Extractions were performed using NE-PER TM Nuclear and Cytoplasmic Extraction kit (Thermo Fisher Scientific #78833) in accordance with the manufacturer's instructions. After extraction, proteins were analyzed by Western blotting as described above.

Statistical analysis
Statistical analyses were performed with Prism 7.0 software (GraphPad). P values lower than 0.05 were considered to be significant. Statistical methods used in the analysis were described in figure legends.