β-Catenin is a pH sensor with decreased stability at higher intracellular pH

β-Catenin functions as both an adherens junction protein for cell–cell adhesion and a signaling molecule for canonical Wnt signaling (Valenta et al., 2012). β-Catenin is essential for growth, differentiation, and cell polarity, and increased β-catenin protein levels, usually achieved by mutation, correlate with pathological progression in epithelial carcinomas (Klaus and Birchmeier, 2008). β-Catenin abundance is predominantly regulated by obligatory phosphorylation of N-terminal residues followed by ubiquitination and proteasome-mediated degradation (Aberle et al., 1997; Daugherty and Gottardi, 2007), although lysosome-mediated degradation (Petherick et al., 2013) and exosome-mediated release (Chairoungdua et al., 2010) have also been reported. The importance of proteasome-mediated degradation in regulating β-catenin function is reflected in cancer-associated mutations that prevent or block proteasome-mediated degradation, resulting in enhanced tumorigenesis (Morin et al., 1997; Peifer and Polakis, 2000).

Proteasome-mediated degradation of β-catenin requires the N-terminal regulatory region (residues 1–133) with obligate phosphorylation of one serine by the priming kinase casein kinase 1 (CK1) followed by processive phosphorylation of two serines and one threonine by glycogen synthase kinase-3β (GSK3β; Liu et al., 2002; Sadot et al., 2002). The E3 ligase β-TrCP recognizes β-catenin at these phosphorylated residues and ubiquitinates β-catenin, targeting it for proteasome-mediated degradation. Mutations in conserved phosphorylated residues Ser33/37 and Thr41 increase β-catenin abundance and correlate with pathological progression in lung (Li et al., 2013), colorectal (Morin et al., 1997), and hepatocellular (Endo et al., 2000) carcinomas. The current view is that phosphorylation of β-catenin by GSK3β at just two residues (Ser33 and Ser37) is both necessary and sufficient for β-TrCP association (Aberle et al., 1997; Ha et al., 2004).

In this study, we report that β-catenin abundance and stability are also regulated by intracellular pH (pHi) dynamics, with increased β-TrCP binding and decreased stability at higher pHi. While β-catenin phosphorylation by both CK1 and GSK3β are unaffected by pHi, an evolutionarily conserved histidine (His36 in human β-catenin) in the β-TrCP binding motif (DSGIHS) mediates pH-sensitive association with β-TrCP. Our data identify pHi dynamics as a previously unrecognized regulator of β-catenin stability, which functions in coincidence with phosphorylation.

We previously reported that overexpression of DNhe2, an orthologue of the mammalian Na–H exchanger NHE1, in the developing Drosophila melanogaster eye increases pHi from 7.3 to 7.7 and is sufficient to induce a rough eye phenotype with underlying dysplasia in the absence of an activated oncogene (Grillo-Hill et al., 2015). To identify potential mediators of the dysplasia phenotype, we performed a dominant modifier screen that revealed a strong genetic interaction between armadillo (arm), the Drosophila β-catenin homologue, and overexpression of DNhe2 under the GMR-GAL4 driver (GMR>DNhe2). Overexpression of arm alone had minimal effects on retinal patterning, causing occasionally misplaced bristles (Fig. 1 A) but suppressed the DNhe2-induced rough eye. Overexpression of arm with DNhe2 restored the hexagonal shape and arrangement of orderly rows of ommatidia (Fig. 1 A). We also found that RNAi-mediated knockdown of arm caused a mild rough eye phenotype with a slightly overgrown appearance (Figs. 1 A and S1 A). This phenotype was enhanced with coexpression of DNhe2 such that eyes were markedly overgrown and showed distinct black granules resembling necrotic granules. These genetic interactions suggest that the rough eye phenotype seen with DNhe2 overexpression may be dependent on decreased Arm protein abundance.

We confirmed decreased Arm abundance with increased pHi by immunoblotting adult Drosophila whole-head lysates. Compared with WT flies, we found lower levels of endogenous Arm with overexpression of DNhe2 but not with a transport-inactive mutant DNhe2^E358I, which does not change pHi or induce a rough eye phenotype (Fig. 1, B and C; Grillo-Hill et al., 2015). These findings suggest that decreased Arm abundance with DNhe2 overexpression is due to the increased pHi and not to scaffolding functions that are retained by the inactive mutant (Denker et al., 2000; Baumgartner et al., 2004). In addition to decreased total Arm abundance, we observed reduced Arm fluorescence intensity at adherens junctions in GMR>DNhe2 eyes compared with control (Fig. 1, D and E).

We next asked whether another adherens junctions protein can rescue the GMR>DNhe2 phenotype. We overexpressed two different transgenes encoding shotgun (shg), the Drosophila homologue of E-cadherin, but neither showed a genetic interaction with DNhe2 overexpression (Fig. S1 B). Consistent with these results, we also observed no difference in the abundance of endogenous Shg or N-cadherin (CadN) across genotypes (Fig. S1, C and D). These data suggest that increased pHi specifically reduces the abundance of Arm instead of reducing the total abundance of adherens junctions proteins in general.

In Drosophila pupal eyes, cadherins are localized to specific cell–cell junctions, with CadN exclusively at cone cell–cone cell contacts and Shg at all other cell–cell contacts. In GMR>DNhe2 pupal eyes, CadN was significantly decreased at cone cell–cone cell contacts (Fig. S1, E and F). However, Shg was significantly decreased at cone cell–primary, primary–primary, and secondary–bristle contacts but not at primary–secondary or secondary–tertiary contacts (Fig. S2, G and H). The contrast with total head lysates, which showed no decrease in Shg or CadN, suggests that changes in junctional localization are independent of total Shg and CadN levels. Taken together, these results indicate that with increased pHi, junctional levels of Arm are reduced across all junctions measured, but effects on junctional cadherin levels are dependent on cell type. The decrease in Arm protein at junctions and in total head lysates suggests a strong and specific pH-sensitive effect on Arm abundance.

Increasing pHi in mammalian cells also reduced β-catenin abundance. We used two methods to increase steady-state pHi in MDCK epithelial cells. Using one previously described approach (White et al., 2017a), we incubated MDCK cells for 24 h with 5 mM NH₄Cl, which increased pHi from 7.4 to 7.8 (Fig. 2 A). Using another approach (Choi et al., 2013), we maintained MDCK cells for 10 d in 15% CO₂, which changes cellular buffering and increased pHi to 7.7 (Fig. 2 A). With both approaches, there was decreased abundance of β-catenin in whole-cell lysates with increased pHi (Fig. 2, B and C). Increasing pHi also decreased β-catenin abundance in mouse mammary NMuMG epithelial cells (Fig. S2, A–C). Additionally, increasing pHi in MDCK cells with NH₄Cl decreased β-catenin abundance at cell–cell junctions and in the nucleus (Fig. 2, D and E). Increasing pHi with 15% CO₂ also decreased β-catenin abundance at junctions but increased the number of double-nucleated cells, suggesting possible cytokinesis defects using this method of pHi control (Fig. S2, D and E). For this reason, we used 5 mM NH₄Cl to increase pHi in all future experiments.

To test whether pHi effects in MDCK are limited to β-catenin or affect other junctional proteins, we measured total E-cadherin levels by immunoblotting, which were unchanged with increased pHi (Fig. S2, F and G). However, we observed a significant increase in E-cadherin localization at cell–cell junctions with increased pHi in MDCK cells (Fig. S2, H and I). Therefore, while we observed differential effects of increased pHi on cadherin localization in Drosophila and mammalian models, Arm/β-catenin reduction at junctions and in whole-cell lysates is universal with increased pHi.

We next asked what effect decreased pHi had on β-catenin abundance. Despite using several approaches, we were unable to significantly decrease pHi in MDCK cells. Therefore, we used NHE1-deficient PS120 fibroblasts stably expressing either WT NHE1 or a mutant (NHE1-AxxA3) with decreased ion transport activity and a significantly lower pHi (Fig. S3 A; Webb et al., 2016). With lower pHi, the abundance of β-catenin at membrane protrusions was significantly increased (Fig. S3, B and C). Taken together, these data suggest that pHi dynamics regulates β-catenin abundance with increased pHi decreasing abundance of total, membrane-associated, and nuclear β-catenin.

To further determine pHi-dependent effects on β-catenin abundance, we measured protein stability by immunoprecipitating endogenous β-catenin after metabolic pulse–chase labeling in MDCK cells. In controls, β-catenin protein had a half-life of 1.6 ± 0.4 h, but with increased pHi, β-catenin half-life was significantly decreased to 0.5 ± 0.05 h (Fig. 2 F). These data suggest that decreased β-catenin abundance in lysates, at cell–cell junctions, and in the nucleus is due to decreased stability of β-catenin protein at higher pHi.

β-catenin stability is primarily regulated by ubiquitination and proteasome-mediated degradation. The E3 ligase β-TrCP binds the N terminus of β-catenin through interaction with the phosphorylated GSK3β sites and ubiquitinates β-catenin (Fig. 3 A, red box; Hart et al., 1999). The evolutionarily conserved β-TrCP recognition motif contains an invariant histidine (His36 in human β-catenin) between the two serine residues phosphorylated by GSK3β (Fig. 3 B). Histidines (∼6.5 pKa) are unique among amino acid residues because they can titrate within the narrow physiological pH range and can function as pH-dependent molecular switches (Schönichen et al., 2013). We previously showed molecular mechanisms for pH-dependent histidine switches in regulating the activity of focal adhesion kinase (Choi et al., 2013), cofilin (Frantz et al., 2008), and the sodium–proton exchanger NHE1 (Webb et al., 2016); however, to our knowledge, the functional significance of histidine in the β-TrCP recognition motif of β-catenin has not been reported.

We first tested whether the histidine in the β-TrCP recognition motif might confer pH-dependent phosphorylation of flanking serine residues by GSK3β. However, in vitro kinase assays performed at buffer pH values of 6.9 and 7.5 using the recombinant N terminus of β-catenin (residues 1–133 fused to GST) revealed pH-insensitive phosphorylation by both the priming kinase CK1 and the processive kinase GSK3β (Fig. 3 C). A published crystal structure of β-TrCP complexed with a double-phosphorylated β-catenin peptide (residues 30–44 with pSer33 and pSer37) shows that β-TrCP WD40 repeats interact electrostatically with β-catenin at both GSK3β phosphorylation sites (Ser33 and Ser37) as well as His36 (Fig. 3 D; Wu et al., 2003). We found a pH-dependent association between endogenous β-catenin and β-TrCP in MDCK cells with increased association at higher pHi. With increased pHi, there was more β-TrCP in β-catenin immune complexes (Fig. 3, E and F) and more β-catenin in β-TrCP immune complexes (Fig. S4, A and B). Additionally, in the presence of the proteasome inhibitor MG132, we observed pH-sensitive changes in ubiquitination of β-catenin, with increased ubiquitination at higher pHi (Fig. 3, E and F). In contrast, we found no pH-dependent changes in β-catenin phosphorylation in MDCK cells, in agreement with our in vitro data (Fig. 3, E and F).

With increased pHi, we found decreased β-catenin in total cell lysates, at cell–cell junctions, and in the nucleus. To examine whether β-TrCP associates with cadherin-bound β-catenin, we used two approaches in MDCK cells. With one approach, we isolated surface proteins using ConA-conjugated beads and found decreased membrane-associated β-catenin at increased pHi (Fig. S4, C and D), which is consistent with our immunolabeling data. However, we were unable to detect membrane-associated β-TrCP, and there was no pHi-dependent difference in E-cadherin abundance. In contrast, with our second approach, we found decreased β-catenin and increased β-TrCP association with endogenous E-cadherin immune complexes at increased pHi (Fig. S4, E and F). These data suggest that β-TrCP can associate with β-catenin–E-cadherin complexes. However, β-TrCP has many cellular targets, and its association specifically with junctional β-catenin and E-cadherin needs further investigation.

We also found increased association of β-TrCP and Arm at higher pHi in Drosophila adult head lysates. We generated Drosophila lines overexpressing HA-tagged Arm and found that there was increased β-TrCP abundance in HA immune complexes with overexpression of DNhe2 to increase pHi (Fig. 4, A and B). β-TrCP is one member of a multiprotein destruction complex that also includes APC, Axin, and GSK3β (Wu et al., 2003; Ha et al., 2004; Valenta et al., 2012). Moreover, GSK3β plays a key role in phosphorylating both APC and β-catenin, which is required for efficient β-catenin degradation (Pronobis et al., 2015) and for formation of the destruction complex (Stamos and Weis, 2013). We verified that formation of the destruction complex was independent of pHi, determined by immunoblotting for GSK3β in HA immune complexes (Fig. 4, A and B). β-TrCP has >40 known target proteins (Frescas and Pagano, 2008; Lee and Diehl, 2014), all sharing a conserved destruction recognition motif (DSGXXS), with a subset having a histidine in the second variable position (DSGXHS; Shafique et al., 2016). We therefore tested whether pH-sensitive association of β-catenin and β-TrCP requires a histidine in the destruction site. We monitored the abundance of disks large (DLG), a β-TrCP target lacking a histidine in its destruction recognition motif (DSGPLS), and found that DLG abundance was pH independent in Drosophila head lysates (Fig. 4, C and D). These data all suggest specificity for pH-sensitive interaction between β-TrCP and β-catenin.

We next directly tested whether the conserved histidine residue in the DSGIHS destruction motif of β-catenin mediates the pH-sensitive association with β-TrCP. We performed an ELISA binding assay with recombinant β-catenin N terminus (residues 1–133) that we phosphorylated in vitro with CK1 and GSK3β. All proteins were purified to similar purity (Fig. S5). We found pH-dependent binding of recombinant β-TrCP to WT phosphorylated β-catenin, with increased β-TrCP binding at higher pH (Fig. 5, A and E). However, β-TrCP binding to a His36-alanine (Ala) mutant (a nontitratable, neutral substitution) was pH insensitive (Fig. 5 B), and binding constants were indistinguishable from WT β-catenin at higher pH (Fig. 5 E). Consistent with β-TrCP binding requiring phosphorylated β-catenin, there was negligible binding of β-TrCP to unphosphorylated WT or H36A–β-catenin at both pH values tested (Fig. 5, A and B). β-TrCP binding to a His36-arganine (Arg) mutant (a nontitratable, positively charged substitution) was pH insensitive; however, there was significant β-TrCP binding to nonphosphorylated H36R–β-catenin (Fig. 5 C). Previous work showed that the N terminus of β-catenin undergoes distinct conformational changes upon phosphorylation (Fig. 5 F; Megy et al., 2005). We predicted that either the increased steric occlusion or charge of the Arg mutation might induce a phosphorylated-like splayed structure even in the nonphosphorylated protein, leading to increased β-TrCP binding to nonphosphorylated H36R–β-catenin. Arg is both bulky and charged, so we also tested a phenylalanine (nontitratable, bulky, and neutral) substitution and observed pH-insensitive β-TrCP binding and significant β-TrCP binding to nonphosphorylated H36F–β-catenin (Fig. 5 D). These data suggest that steric effects are important for specific recognition of phosphorylated β-catenin over nonphosphorylated β-catenin (Fig. 5 F) and that both established steric constraints conferred by phosphorylation and charge effects at His36 coincide to regulate β-TrCP recognition of β-catenin.

Somatic mutation of the conserved His36 to Arg has been identified in several cancers in the Catalog of Somatic Mutations in Cancer database (Forbes et al., 2015); however, to our knowledge, the functional consequences of this mutation have not been reported. We generated transgenic Drosophila lines with inducible expression of WT Arm (UAS-arm), with the analogous His42-Arg mutation (UAS-arm^H42R), and an Ala substitution His42-Ala (UAS-arm^H42A). Transgenes were expressed using eyeless-GAL4 (eyGAL4) at moderate levels in retinal tissues from the time of specification in embryos through adulthood. Expression of WT arm showed no patterning defects in the eye (Fig. 6, A and B), and the arm^H42A mutation showed ∼55% WT eyes, ∼37% eyes with mildly rough patches, and ∼8% eyes with slight overgrowth (Fig. 6, A and B). In contrast, all eyes expressing arm^H42R showed profound defects in retinal patterning, with specific defects including reduced rough eyes (∼59%), overgrown bulging eyes (∼19%), reduced split eyes (∼8%), missing eyes (∼5%), and strikingly large protruding tumors (∼8%; Fig. 6, A and B). Histological sections through these tumors revealed mostly well-differentiated photoreceptor neurons, with a few ommatidia missing a full complement of eight photoreceptor neurons (Fig. 6 C, black circles). Similar phenotypes were reported with eyGAL4-expressed transgenes that activate Wnt signaling in the eye, where expression of RNAi targeting slimb (Drosophila β-TrCP homologue) or shaggy (Drosophila GSK-3β homologue) induced overgrowth or protruding tumors, respectively, while overexpression of disheveled showed the reduced, split-eye phenotype (Reim et al., 2014). In addition, histological sections of tumors induced by arm^H42R revealed significant planar cell polarity errors, including inversions on the dorsal–ventral and anterior–posterior axes (Fig. 6 C); the penetrance of these errors was so high that it was impossible to discern the axes of the eye from these sections. We compared the penetrance of these phenotypes to the classic gain-of-function, stabilized mutant-variant arm^S10 (Pai et al., 1997), where a portion of the N terminus containing the phosphorylated and ubiquitinated residues is deleted. With expression of arm^S10, ∼9% of eyes were mildly rough, ∼67% were reduced and rough, and ∼21% showed overgrown, bulging eyes. The appearance of unique phenotypes in arm^H42R that are not seen with arm^S10 suggests that this mutation may act as a neomorph, conferring novel protein functions.

To determine whether these mutations affect canonical Wnt signaling, we generated flies expressing our UAS-arm transgenes under control of eyGAL4 with the established transcriptional reporter notum-lacZ (Liu et al., 2008). Reporter expression in third larval instar eye imaginal discs was examined using X-gal staining. In control crosses with the WT strain w¹¹¹⁸, notum-lacZ expression is restricted to the dorsal and ventral poles of the eye imaginal disc in two distinct patches (Fig. 6 D). Flies with ey>arm^WT or ey>arm^H42A express these transgenes throughout the eye disc but showed a similar pattern to control, suggesting that endogenous degradation mechanisms are sufficient to prevent ectopic Wnt signaling for these transgenes (Fig. 6 D). However, flies expressing ey>arm^H42R showed elevated Wnt activity throughout the eye disc in all samples examined (Fig. 6 D). These data indicate that when the arm^H42R mutant is expressed at physiologically relevant levels, it is both resistant to endogenous degradation mechanisms and sufficient to induce aberrant Wnt signaling.

Our work identifies pHi dynamics as a previously unrecognized regulator of β-catenin stability. We show that β-catenin is a pH sensor, and at higher pHi, it has decreased abundance in whole lysates, at adherens junctions, and in the nucleus. We suggest a neutral (deprotonated) histidine in the destruction motif DSGXHS functions in coincidence with obligatory phosphorylation of N-terminal residues (Aberle et al., 1997; Daugherty and Gottardi, 2007) to enable β-catenin binding to the E3 ligase β-TrCP. We previously reported that cofilin, an actin-binding protein and pH sensor, also requires a neutral histidine (His133 human cofilin) in coincidence with an obligatory N-terminal phosphorylation (pSer3) for increased activity (Frantz et al., 2008). The DSGXHS degradation motif with an invariant histidine in β-catenin also occurs in a number of other proteins, and an important future direction is to determine whether the invariant histidine confers pHi-dependent changes in stability in those proteins as well. Our work reveals an unexplored effect of pHi dynamics on proteasome-mediated protein stability: if this pHi-dependent regulation is common to a subset of β-TrCP targets, it may enable coordinated cellular responses to increased pHi.

A number of normal cell behaviors regulated by pHi dynamics including migration (Denker and Barber, 2002; Choi et al., 2010; Parks and Pouyssegur, 2015), differentiation (Ulmschneider et al., 2016), and tumorigenesis (Reshkin et al., 2014; White et al., 2017b) are also regulated by Wnt signaling (Krausova and Korinek, 2014; Sedgwick and D’Souza-Schorey, 2016). For example, in follicle stem cells of the Drosophila germarium, Wnt signaling is high and decreases in differentiated daughter follicle cells (Sahai-Hernandez et al., 2012; Dai et al., 2017; Waghmare and Page-McCaw, 2018), and we recently showed that pHi is lower in follicle stem cells, with an obligatory increase for differentiation of follicle cells (Ulmschneider et al., 2016). Future work testing these collective findings will be important for understanding how pHi dynamics cross-talks with both Wnt signaling and other signaling pathways in normal and pathological cellular processes.

The regulation of β-catenin stability by pHi dynamics likely has previously unrecognized roles in cancer. Most cancers have a constitutively increased pHi >7.4 (White et al., 2017b), and increased pHi is sufficient to induce dysplasia in the absence of activated oncogenes (Grillo-Hill et al., 2015). Relevant to cancer, our findings using the Drosophila transcriptional reporter notum-LacZ suggest that an H36R–β-catenin somatic mutation in human cancers (Forbes et al., 2017) can bypass pH-sensitive β-TrCP recognition and increase Wnt pathway activity. Past work by Simons et al. (2009) showed that increased pHi induced by overexpression of a Drosophila Na–H exchanger promotes noncanonical Wnt signaling; however, no discernable effects on canonical Wnt signaling were found. Our work adds to the growing body of data suggesting pHi as a regulator of diverse cellular signaling processes and supports future studies determining whether pH-dependent cancer cell behaviors are in part mediated by pH-dependent changes in β-catenin stability such as loss of cell–cell junctions for tumor cell metastasis.

Flies were grown on standard Drosophila media and cultured at 25°C. The following lines were used: GMRGAL4 (from R. Cagan, Mount Sinai School of Medicine, New York, NY); yw; Notum-lacZ; (from K. Cadigan, University of Michigan, Ann Arbor, MI); w; UAS-DNhe2 (described by Grillo-Hill et al., 2015); y[1] w[1118]; P{w[+mC]=UAS-arm.S2}A1 (UAS-arm; 4783; Bloomington Drosophila Stock Center [BDSC]); y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.JF01251}attP2/TM3, Ser[1] (UAS-arm.RNAi4; 31304; BDSC); y[1] v[1]; P{y[+t7.7] v[+t1.8]=TRiP.JF01252}attP2 (UAS-arm.RNAi5; 31305; BDSC); w[1118]; P{w[+mC]=UASp-shg.GFP}5B (UAS-shg::GFP; 58445; BDSC); w[1118]; P{y[+t7.7] w[+mC]=UASp-shg.S}attP2 (UAS-shg; 65589; BDSC); P{w[+mC]=UAS-arm.S10}C, y[1] w[1118] (UAS-Arm.S104782; BDSC); y[1] w[1118]; P{w[+mC]=ey3.5-GAL4.Exel}3, (eyGAL4; 8219; BDSC); and pnut[XP]/T(2;3)SM6a-TM6B, Tb[1] (SM6a-TM6B; 5687; BDSC).

The arm cDNA was obtained from a public stock center (cDNA clone LD23131; Drosophila Genomics Resource Center) and cloned into pUASt-HA to generate an HA tag at the C terminus of the protein to distinguish endogenous from transgenic protein. Transgenic flies were generated (BestGene, Inc.) and phenotypically tested, and protein expression levels were empirically determined. For these studies, we selected lines with high and low protein expression levels and saw identical results for both.

Adult heads were removed, bisected sagittally, and fixed overnight in 2% glutaraldehyde + 4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4. Samples were processed in a critical point dryer, mounted on carbon tape (16,084-6; Ted Pella) or with colloidal silver (16,034; Ted Pella) on an scanning EM stub (16,221; Ted Pella) and splutter coated with gold-palladium (22-2; Ted Pella). Images were collected on a JEOL Neoscope scanning electron microscope and cropped in Photoshop (Adobe).

For Drosophila immunolabeling, the following antibodies were obtained from the Developmental Studies Hybridoma Bank (DSHB; created by the National Institute of Child Health and Human Development of the National Institutes of Health and maintained at The University of Iowa): Arm (N2 7A1; 1:10; deposited to the DSHB by E. Wieschaus, Princeton University, Princeton, NJ), Shg (DCAD2; 1:10; deposited to the DSHB by T. Uemura, Kyoto University, Kyoto, Japan), and N_cadherin (DN-Ex 8; 1:10; deposited to the DSHB by T. Uemura). For MDCK immunolabeling, the following antibodies were used: mouse anti–β-catenin (1:400; 610154; BD) and rat anti–E-cadherin (1:400; 13-1900; Invitrogen).

Drosophila pupal retinal tissues were dissected in PBS, fixed in fresh 4% paraformaldehyde for 30 min, washed in PBS + 0.1% Triton X-100, and immunostained with primary antibodies overnight at 4°C and secondary antibodies for 60 min at RT in the dark. Mammalian cells were grown as described below, fixed on ice for 20 min in fresh 4% paraformaldehyde, washed in PBS, blocked in PBS + 5% goat serum + 0.3% Triton X-100 for 15 min at RT, and washed in PBS. Fixed cells were immunostained with primary antibodies overnight at 4°C and secondary antibodies for 60 min at RT in the dark. Samples were mounted in Prolong Gold (P36930; Thermo Fisher Scientific). Confocal images were collected on an inverted spinning-disk confocal microscope (Stehbens et al., 2012). NIS Elements software (Nikon) was used to subtract background and measure fluorescence intensity in individual channels. For display purposes, single-channel images were exported, and the lookup table “Fire” was applied in FIJI (ImageJ; National Institutes of Health). For intensity measurements, fluorescence intensity measurements were collected from a single region of interest per cell–cell junction in the Z section with highest intensity for that junction. Statistical analyses were performed using Prism (GraphPad Software). First, data were subjected to the D’Agostino and Pearson omnibus normality test, and all datasets showed a nonnormal distribution. Outliers were removed using the robust regression and outlier removal (ROUT) method. Where indicated, data were next normalized within each biological replicate to the mean intensity for all control junctions. Data were then subjected to the Mann–Whitney nonparametric test to determine significance.

Adult eye tissue dissection, fixation, and resin embedding were performed as described previously (Wolff, 2000). Tissue was cut into 0.5-µm sections on a microtome and contrast stained with toluidine blue. Phase-contrast images were collected on a Zeiss Axiophot microscope using a 63× oil objective and a Hamamatsu ORCA-ER digital camera. Images were cropped, pseudocolored, and brightness/contrast adjusted in Photoshop.

For Wnt reporter assays, flies expressing UAS-arm transgenes under the eyeless-GAL4 driver were generated and balanced over the SM6a-TM6B compound balancer. Genotypes were verified by PCR, where both wings were removed from individual male flies and used as the source for the genomic DNA. Males carrying both the GAL4 and UAS transgenes balanced over SM6a-TM6B were mated to notum-lacZ virgin females. Resulting offspring were grown to wandering third instar larvae and then dissected, and eye imaginal discs were fixed in 1% glutaraldehyde for 15 min at RT and permeabilized in 0.3% Triton X-100 in PBS. X-gal staining was performed as described previously (Cold Spring Harbor Protocols, 2010). Tissues were mounted in glycerol and imaged as described above.

Drosophila heads were dissected into cold radioimmunoprecipitation assay buffer (Peifer, 1993; 300 mM NaCl, 50 mM Tris-HCl, pH 8.5, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 2 mM CaCl₂, 1% Triton X-100, 1% NP-40 in PBS plus 10 mM DTT, 1 mg/ml PMSF, 10 mM Na₃VO₄, 10 mM NaF, and EDTA-free complete protease inhibitors [Roche]) at 2–3 µl per head. Heads were homogenized with a pestle, incubated for 60 min on ice, and then sonicated on ice for six rounds of 30-s sonication with 90 s rest on ice. Lysates were centrifuged for 15 min at 4°C at 15,000 g in an Eppendorf centrifuge. For each condition, 10 heads were loaded onto 10% SDS-PAGE gels and run at 120 V for 1.2 h. Proteins were transferred (25 mM Tris, 192 mM glycine, 20% methanol, and 0.01% SDS) onto polyvinylidene difluoride (PVDF) membranes at 100 V for 1.5 h. Membranes were blocked for 1 h in 5% nonfat milk in TBS-T (TBS with 0.1% Tween-20) with shaking at RT. Primary antibodies were diluted in 5% nonfat milk in TBS-T and incubated overnight with shaking at 4°C. Arm (1:500; N2 7A1), Shg (1:1,000; DCAD2), N_cadherin (1:1,000; DN-Ex 8), DLG (1:1,000; 4F3), and actin (1:5,000; C4; EMD Millipore). Membranes were washed in TBS-T (3 × 5 min) with shaking. Appropriate secondary antibodies were diluted 1:10,000 in 5% nonfat milk in TBS-T and incubated for 1 h with shaking at RT (Bio-Rad anti–mouse-HRP [170-6516] and anti–rabbit-HRP [172-1019]). Membranes were washed in TBS-T (3 × 5 min) with shaking, developed (SuperSignal West Femto; Pierce), and visualized (Alpha Innotech FluorChem Q).

MDCK (ATCC): Eagle’s minimum essential medium + 10% FBS. NMuMG cells: DMEM supplemented with 10% FBS and 10 µg/ml insulin. PS120 cells: DMEM supplemented with 5% FBS. MDCK, NMuMG, or PS120 cells were plated (6.0 × 10⁴ cells per six-well plate for MDCK, 4.0 × 10⁴ for NMuMG, or 5.0 × 10⁴ for PS120). For acute pHi control, cells were incubated in complete medium supplemented with 5 mM ammonium chloride for 24 h. For chronic pHi control, MDCK cells were placed in an incubator with atmospheric conditions of 15% CO₂ for two passages (10 d) and increased HCO₃ in the media to 30 mM. Where indicated, the proteasome inhibitor MG132 (S2619; SelleckChem) was added to a final concentration of 10 µM for the final 6 h of the experiment. Steady-state pHi was measured as previously described (Choi et al., 2013). For all conditions, triplicate wells were plated (2 × 10⁴ per well; 24-well plate). Technical replicates were averaged within each biological replicate.

Cells were washed twice with ice-cold PBS and incubated with 100 ml ice-cold lysis buffer (50 mM Tris, 150 mM NaCl, 1 mM NaF, 1% Triton X-100, and protease inhibitor cocktail [Roche], pH 7.5) for 10 min on ice. Cells were scraped, and clarified lysate (10,000 rpm; 10 min) was used immediately for immunoprecipitation (IP), immunoblotting, or flash freezing and storage at −80°C.

Proteins were transferred to 0.45-µm PVDF membranes (100 V for 1 h on ice; Immobilon) in modified transfer buffer (Tris-Gly, 20% MeOH, and 0.01% SDS) to maximize transfer of ubiquitinated β-catenin. Membranes were blocked in 5% BSA in TBS-T (0.1% Tween-20 in TBS) for 1 h at RT and then divided for blotting. Primary antibodies: total β-catenin (1:1,000; 610154; BD), pSer (33/37) pThr41 β-catenin (1:1,000; 9561; Cell Signaling Technology), pSer45 β-catenin (1:1,000; 9564; Cell Signaling Technology), and β-TrCP (1:1,000; D13F10; 4349; Cell Signaling Technology), ubiquitin (1:2,000; 3933; Cell Signaling Technology), and E-cadherin (1:2,000; 610181; BD). Primary antibodies were used at indicated dilutions in 5% BSA/TBS-T and incubated overnight with shaking at 4°C. Membranes were washed with shaking (3 × 5 min; TBS-T). Secondary HRP-conjugated antibodies against mouse or rabbit (1706515 and 1721019; Bio-Rad; 1:10,000 in 5% BSA/TBS-T) were incubated for 1 h (RT with shaking). Membranes were washed with shaking (3 × 5 min; TBS-T), developed (SuperSignal West Femto; Pierce), and visualized (Alpha Innotech FluorChem Q).

Cell lysates were prepared as described above. 250 µl lysate was combined with 1–2 µl anti–β-catenin antibody (AB19022 [discontinued; EMD Millipore] or 9562 [Cell Signaling Technology] for β-TrCP coimmunoprecipitations; 610154 [BD] for ³⁵S metabolic pulse–chase assays), anti–β-TrCP antibody (D13F10; 4394; Cell Signaling Technology), or anti–E-cadherin antibody (610181; BD). Antibody and lysate were incubated on a nutator overnight at 4°C. After overnight incubation, 30–40 µl protein G-Mag Sepharose slurry (28-9440-08; GE Healthcare) was added to each tube and incubated on a nutator for 3 h at 4°C. Protein G-Mag Sepharose beads were separated by magnet, and supernatant was removed. Beads were washed 3× with 800 µl wash buffer (TBS + 0.1% Triton X-100 and protease inhibitor cocktail). Protein was eluted from washed beads by incubating at 50°C for 5 min with 40 µl 2× loading buffer (Bio-Rad) without DTT. Beads were separated by magnet, elution was removed, and DTT was then added to a final concentration of 100 mM. For ConA separations, 30 µl ConA-conjugated agarose (786-216; G-biosciences) was added to lysates and incubated overnight on a nutator at 4°C. Washes were all performed as described above, spinning agarose between washes at 500 rpm for 2 min at 4°C. For all conditions, 5–10 µl of the elutions was run on 10% SDS-PAGE gels. For representative blots in Fig. 3, equal amounts of β-catenin protein loading were achieved by adding 50% more elution in the high-pH conditions. Immunoblotting was performed as described above.

MDCK cells were plated at 8.0 × 10⁴ cells per 35-mM dish with eight total dishes plated. 24 h after plating, half of the dishes were treated with 5 mM ammonium chloride. 48 h after plating, cells were starved of methionine by washing and incubating for 30 min in methionine-free media (Eagle’s minimum essential medium without methionine and cysteine [1641454; MB-bio] supplemented with 10 mM cysteine, 2 mM glutamine, and 10% FBS). Note that for this and all future steps, 5 mM ammonium chloride was supplemented to media and washes for the high-pHi conditions. After methionine depletion, cells were incubated with 0.5 ml methionine-free media supplemented with 0.5–0.625 mCi/ml ³⁵S-methionine (NEG009A; PerkinElmer) for 30 min. Cells were washed 3× in complete media, and time-0 lysates were prepared as described above for control and ammonium chloride conditions. All other plates were returned to the incubator, and matched plates were lysed 1 h, 2 h, and 4 h later. For two of the four replicates, the experiment was performed as described above except using double-labeled ³⁵S-methionine/³⁵S-cysteine (NEG772; PerkinElmer) supplemented at 0.5–0.875 mCi/ml. For these conditions, cells were starved of methionine and cysteine by washing and incubating for 30 min in methionine/cysteine-free media (Eagle’s minimum essential medium without methionine and cysteine [1641454; MB-bio] supplemented with 2 mM glutamine and 10% FBS).

250 µl of each labeled lysate was combined with 1 µl anti–β-catenin antibody (610154; BD). Antibody and lysate were incubated on a nutator overnight at 4°C. After overnight incubation, 30–40 µl protein G-Mag Sepharose slurry (28-9440-08; GE Healthcare) was added to each tube and incubated on a nutator for 3 h at 4°C. Protein G-Mag Sepharose beads were washed as described above, and protein was eluted from washed beads by boiling at 95°C in 25 µl 2× Laemmli buffer with DTT. The entire elution was run on a 10% SDS-PAGE gel, which was then stained for 1 h in Instant Blue (1SB1L; Expedeon). After destaining, an image was acquired, and then gels were fixed (10% acetic acid and 40% methanol in water) for 5 min, rehydrated in 2% glycerol in water for 5 min, and then incubated in FluorHance (Research Products International) for 30 min before drying (model 853; Bio-Rad). Autoradiograph film (E3018; Denville Hyblot CL) was exposed for 1–2 wk at −80°C. Densitometry results from the autoradiograph were normalized to amount of protein detected in IP by InstantBlue staining.

Drosophila heads were harvested and prepared as described above. For each IP sample, clarified lysate from 100 heads were combined with 2 µl anti-HA antibody (HA.11; 16B12; Covance), and IPs were performed as described above for mammalian cells. Elution volume for each IP sample was 30 µl, and 15 µl was run on 10% SDS-PAGE gels. Immunoblotting was performed as described above using anti-Arm (1:500; N2 7A1; DSHB), anti–β-TrCP (1:200; D13F10; 4394; Cell Signaling Technology), and anti-GSK3β (1:2,000; 05-412; EMD Millipore).

Original WT β-catenin (residues 1–133 fused to GST) was generously provided by B. Weis (Stanford University, Stanford, CA). Point mutations (H36A, H36R, and H36F) were introduced using standard QuikChange protocol (Stratagene). Resulting constructs were transformed into and expressed in BL21-DE3 Escherichia coli. Cells were grown in Luria broth + 50 µg/ml ampicillin (37°C; with shaking) until OD₆₀₀ = 0.6 and induced (0.5 mM IPTG for 5 h at RT with shaking). Cells were pelleted (6,000 g for 10 min at 4°C), resuspended in lysis buffer (25 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, and protease inhibitor cocktail), and lysed by sonicating 6 × 30 s (cooling lysate for 2 min between sonication pulses). Clarified supernatant (14,000 g for 10 min at 4°C) was loaded onto to a preequilibrated glutathione-agarose column (Thermo Fisher Scientific) and washed with 20 column volumes lysis buffer, and then the protein was eluted in binding buffer with 15 mM reduced glutathione, pH 8.0. Proteins were dialyzed into storage buffer (20 mM Tris, 50 mM NaCl, 1 mM DTT, and 5% glycerol, pH 7.5) and concentrated. Aliquots were flash frozen in liquid nitrogen and stored at −80°C.

For kinase reactions, 15 µg purified β-catenin–GST was incubated with 0.4 µg recombinant CK1 (SRP5013; Sigma-Aldrich) either with or without 0.2 µg recombinant GSK3β (G4296; Sigma-Aldrich). Reactions were performed in 50 mM Hepes buffer containing 5 mM ATP, 10 mM MgCl₂, and 5 mM DTT, pH 6.9 or 7.5, for 1 h at 30°C. Reactions were quenched with EDTA (C_f = 100 mM). Each reaction was boiled in 1× Laemmli buffer for 5 min and then divided in half and run in duplicate on 12% SDS-PAGE gel at 100 V for 90 min. Western blotting for phospho–β-catenin was performed as described above.

30 µg purified protein was phosphorylated as described above at buffer pH 7.5 for 4 h to ensure complete phosphorylation. Phosphorylated or nonphosphorylated β-catenin was immobilized on high-binding ELISA plates at 4 µg/ml in PBS overnight with shaking at 4°C. Plates with immobilized protein were washed 2× with PBS and then blocked with 5% milk in PBS overnight with shaking at 4°C. Blocked plates were washed 2× with PBS, and then WT β-TrCP was incubated at concentrations from 0–1 µM in binding buffer (20 mM Hepes, 150 mM NaCl, and 1% goat serum) at pH 6.8 or 7.5 for 90 min at RT with shaking. Plates were washed with wash buffer (binding buffer + 0.05% Tween-20) 3 × 5 min and then incubated for 2 h with anti–β-TrCP (D13F10; 4394; Cell Signaling Technology) diluted 1:1,000 in binding buffer. Plates were washed with wash buffer for 3 × 5 min and then incubated for 90 min with anti–rabbit-HRP (1721019; Bio-Rad) at 1:5,000 dilution. Plates were washed with wash buffer for 3 × 5 min, 100 µl 3,3′,5,5′-tetramethylbenzidine (T8665; Sigma-Aldrich) was added to each well for 15 min, and the optical density was read at 370 nM on a SpectraMax M5 plate reader.

Statistical analyses used to compare datasets are described in the figure legends. N indicates number of biological replicates, while n indicates number of technical replicates (or individual measures where applicable). Where appropriate and where indicated, medians were displayed instead of means. Statistical tests used for normality testing (where applicable) are described in Materials and methods.

Fig. S1 shows that overexpression of DNhe2 has no effect on total cadherins and mixed effects on cadherin localization at cell–cell contacts. Fig. S2 shows that higher pHi decreases β-catenin total and junctional abundance in NMuMG epithelial cells and in MDCKs (15% CO₂); while total E-cadherin levels are unchanged with altered pHi, junctional levels in MDCKs are altered. Fig. S3 shows that lower pHi increases β-catenin abundance at membrane protrusions in fibroblasts. Fig. S4 shows that higher pHi increases association of β-catenin with the E3 ligase β-TrCP and increases the amount of β-TrCp associating with cadherin-associated β-catenin. Fig. S5 shows a Coomassie stain of purified β-catenin–(1–133)-GST proteins.

We thank Changhoon Choi for assistance on the early stages of the project, Marian Baquilar for assistance with screening and genetic interactions, and Luke Miller, Bill Weis, James Nelson, Torsten Wittmann, and members of the Wittmann and Barber laboratories for helpful discussions. Drosophila fly lines were obtained from the Drosophila Genomics Resource Center, supported by National Institutes of Health grant 2P40OD010949, and plasmids for in vitro β-catenin–(1–133)-GST were provided by B. Weis.

This work was supported by National Institutes of Health grants CA197855 and GM116384 to D.L. Barber, National Institutes of Health F32 grant CA177085 to K.A. White, and startup funding from the San Jose State University Department of Biological Sciences and College of Sciences to B.K. Grillo-Hill.

The authors declare no competing financial interests.

Author contributions: K.A. White and B.K. Grillo-Hill contributed equally to this work. B.K. Grillo-Hill prepared all Drosophila reagents and performed screening of genetic interactions, scanning EM, histology, immunolabeling, and image analysis, and they also prepared fly head lysates. K.A. White performed biochemical and cell biology analyses, metabolic pulse–chase labeling, pH measurements, and image analysis. M. Esquivel, J. Peralta, V.N. Bui, and I. Chire screened for genetic interactions. J. Peralta, V.N. Bui, and I. Chire assisted with the Wnt reporter assays in flies. D.L. Barber performed pH measurements, immunolabeling, metabolic pulse–chase labeling, and image analysis. D.L. Barber, K.A. White, and B.K. Grillo-Hill conceived experiments, performed data analysis, and wrote the manuscript. D.L. Barber should be contacted for any cell biological or biochemical questions/reagents. B.K. Grillo-Hill should be contacted for any Drosophila questions/reagents.