Gαi2-induced conductin/axin2 condensates inhibit Wnt/β-catenin signaling and suppress cancer growth

Abstract

Conductin/axin2 is a scaffold protein negatively regulating the pro-proliferative Wnt/β-catenin signaling pathway. Accumulation of scaffold proteins in condensates frequently increases their activity, but whether condensation contributes to Wnt pathway inhibition by conductin remains unclear. Here, we show that the Gαi2 subunit of trimeric G-proteins induces conductin condensation by targeting a polymerization-inhibiting aggregon in its RGS domain, thereby promoting conductin-mediated β-catenin degradation. Consistently, transient Gαi2 expression inhibited, whereas knockdown activated Wnt signaling via conductin. Colorectal cancers appear to evade Gαi2-induced Wnt pathway suppression by decreased Gαi2 expression and inactivating mutations, associated with shorter patient survival. Notably, the Gαi2-activating drug guanabenz inhibited Wnt signaling via conductin, consequently reducing colorectal cancer growth in vitro and in mouse models. In summary, we demonstrate Wnt pathway inhibition via Gαi2-triggered conductin condensation, suggesting a tumor suppressor function for Gαi2 in colorectal cancer, and pointing to the FDA-approved drug guanabenz for targeted cancer therapy.

Fig. 1. Gαi2 induces conductin condensates.

a Structural alignment of the modeled conductin RGS domain (black) to RGS1 (purple) co-crystalized with Gαi1 (gray) (PDB ID: 2GTP). The conductin aggregon and its two key residues (QV) are highlighted in blue and orange, respectively. b, c Immunofluorescence staining of Flag (red) in U2OS cells transfected with Flag-tagged conductin (Cdt), its M3 mutant, and different GFP-tagged Gα proteins (green) either alone (b) or in combinations (c), as indicated. For co-expression of conductin with Gαi2, examples of cells with condensate formation (upper row) and membrane recruitment (lower row) are shown. Insets are magnified in the lower right corner. Arrowheads point to membrane recruitment. Scale bars: 20 µm. d, e Percentage of 900 transfected cells out of three independent experiments as in (c) that show condensation (d) or membrane recruitment (e) of conductin or the M3 mutant. Results are mean ± SEM (n = 3). **p < 0.01, ***p < 0.001 (two-sided Student’s t-test). f Western blotting for GFP, Flag, and β-actin (loading control) in lysates of U2OS cells transfected with indicated constructs. One out of three representative experiments is shown. Molecular weight is indicated in kDa. Source data and exact p-values are provided as source data file.

Fig. 2. Gαi2 induces conductin polymerization via steric interference with RGS–RGS aggregation.

a Western blotting for GFP and HA in lysates (L) of HEK293T cells transfected with indicated constructs, and after GST-pulldown from these lysates. b Quantification of the RGS domain amounts that were co-precipitated with the indicated Gα proteins in six independent experiments as in (a). RGS amounts are normalized to the precipitated amounts of the Gα proteins and presented relative to Gαo. c Quantification of RGS domain binding as described in (b) from five independent experiments as shown in Supplementary Fig. 4b. d, e Percentage of 900 transfected cells out of three independent experiments similarly performed as in Fig. 1c that show condensation (d) or membrane recruitment (e) of conductin. f Relative binding of purified GST (control) and GST-tagged conductin fragments containing the RGS domain (GST-RGS^Cdt, GST-Cdt 2–210) to purified Gαi2 preloaded with GDP, GDP + AlF₄⁻, or GTPγS, quantified from three independent pulldown experiments with purified recombinant proteins, as shown in Supplementary Fig. 4e. g Immunofluorescence staining of HA-Conductin (Cdt) in HEK293 shRNA-control cells which were untreated or treated with 30 µM AlF₄⁻ for 1 h. Scale bar: 10 µm. h Percentage of 1500 transfected cells out of five independent experiments as in (g) that show condensation of HA-Conductin. The cells stably express either a control shRNA (−) or indicated shRNAs against Gα proteins. i, k Western blotting under native and denaturing (denat.) conditions for HA and GFP in lysates of HEK293T cells, which were transfected and treated with AlF₄⁻, as indicated. j Western blot quantification of large, medium, and small aggregation complexes according to indications L, M, and S in (i) as percentage of the total protein amount, from three independent experiments. Results are mean ± SEM (n = 6 [b], n = 5 [c, h], n = 3 [d–f, j]). *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided Student’s t-test). Molecular weight is indicated in kDa (a, i, k). Source data and exact p-values are provided as source data file.

Fig. 3. Induction of conductin polymerization requires aggregon masking.

a Schematic presentation showing DIX domain-mediated conductin polymerization after RGS aggregation mediated by the blue aggregon gets sterically inhibited by Gαi2 binding (left), contrasting the persisting RGS aggregation mediated by the red aggregation site in spite of Gαi2 binding in the QVL mutant (right). GSK3 (G) and β-catenin (β) binding sites are indicated. b Structural alignment of the modeled conductin RGS domain (black) to the structure of RGS1 (purple) co-crystalized with Gαi1 (gray) (PDB ID: 2GTP). The conductin aggregon and the silent aggregation site are highlighted in blue and red, respectively. Arrowheads indicate the accessibility of the aggregation sites without and with Gα protein binding. c Immunofluorescence staining for Flag (red) in U2OS cells transfected with Flag-Conductin QVL together with Gαi2-GFP (green). Arrowheads point to membrane recruitment. Scale bar: 20 µm. d, e Percentage of 1200 transfected cells out of four independent experiments as in (c) that show condensation (d) or membrane recruitment (e) of conductin or the QVL mutant. f Western blotting for GFP and HA in lysates (L) of HEK293T cells transfected with indicated constructs, and after GST-pulldown from these lysates. Molecular weight is indicated in kDa. g Quantification of Gαi2-binding to the WT and QVL-mutated conductin RGS domain from five independent GST-pulldown experiments as in (f). Results are mean ± SEM (n = 4 [d, e], n = 5 [g]). *p < 0.05, ***p < 0.001 (two-sided Student’s t-test). Source data and exact p-values are provided as source data file.

Fig. 4. Gαi2 promotes conductin-mediated inhibition of Wnt signaling.

a Immunofluorescence staining of endogenous β-catenin (red) in SW480 cells, which were transfected and treated as indicated on the left. Scale bar: 20 µm. b Quantification of β-catenin and GFP fluorescence intensity in four independent experiments as in (a). c Western blotting for Flag, HA, and GFP in lysates of HEK293T cells transfected as indicated above the blots. GFP: loading and transfection control. * indicates unspecific band. Quantification of Flag-β-catenin normalized to GFP of three independent experiments. d, f–h Luciferase activity (TOP/FOP) in U2OS cells transfected with indicated siRNAs (d, f), or/and indicated amounts of Gαi2 (g, h). Wnt3a treatment is indicated. e Western blotting showing the efficiency of Gαi2 knockdown in U2OS cells. Consistent with TOP-Flash activation (d), expression of the β-catenin target gene conductin (Cdt) was increased upon knockdown. α-Tubulin: loading control. i, k Luciferase activity (TOP/FOP) in Gαi2 transfected DLD1 cells without (–) and with conductin knockdown (i), and in parental SW480 cells, a WT control clone and two CRISPR/Cas9 AXIN2/Conductin knockout clones (Cdt KO) (k). To facilitate comparisons of the Gαi2 effects in cells with (black bars) and without conductin (siCdt/Cdt KO, red bars) in (h, i, k), the initial luciferase activities without Gαi2 were set to 100% for both conditions, and the luciferase activities with Gαi2 are presented relative to the respective initial activity. j, l Western blotting showing conductin knockdown and Gαi2 expression in the DLD1 cells used in (i, j), and loss of conductin expression in the SW480 knockout clones used in (k, l). α-Tubulin: loading control. Results are mean ± SEM (n = 80 [b], n = 3 [c], n = 5 [d, f], n = 4 [g, h, i, k]). *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided Student’s t-test). Molecular weight is indicated in kDa (c, e, j, l). Source data and exact p-values are provided as source data file.

Fig. 5. Inhibitory Gαi2 aberrations are associated with reduced survival of colon cancer patients.

a Strong negative correlation of GNAI2 mRNA expression with β-catenin target genes (AXIN2, RNF43, ASCL2, LGR5) as shown by the negative Pearson’s correlation coefficients (R) and p-values, and no or positive correlation with housekeeping genes (ACTB, GAPDH) in normal human colon tissue and colorectal cancer analyzed with GEPIA. b Copy number variation (CNV) of the GNAI2 gene observed in the TCGA colorectal cancer data sets COAD and READ. c GNAI2 mRNA expression in tumors (T) of the COAD and READ data set compared to normal colon tissue (N). TPM: transcripts per million. d, e Luciferase activity (TOP/FOP) in DLD1 cells (d) and SW480 WT and AXIN2/Conductin knockout (Cdt KO) cells (e) transfected with two different siRNAs against Gαi2 (A, B). f Three GNAI2 colorectal cancer mutations that were predicted as “deleterious” by Sorting Intolerant From Tolerant (SIFT). g Western blotting for GFP and α-tubulin (loading control) in lysates of HEK293T cells used in (h). Molecular weight is indicated in kDa. h Luciferase activity (TOP/FOP) in Wnt3a-treated HEK293T cells transfected with rising amounts of WT Gαi2-GFP or colorectal cancer mutants (CRC mut). i Survival analysis of colon cancer patients with GNAI2 missense mutations or copy number loss (n = 25) compared to patients without such GNAI2 alterations (n = 338) in the COAD data set. Results are mean ± SEM (n = 4 [d, e, h]). *p < 0.05, **p < 0.01 (two-sided Student’s t-test). Source data and exact p-values are provided as source data file.

Fig. 6. Activation of Gαi2 by GBZ treatment inhibits Wnt signaling in colorectal cancer cells.

a Immunofluorescence staining of Flag-Conductin (Cdt) in U2OS cells, which were treated with 50 µM GBZ overnight. Scale bar: 20 µm. b Percentage of 1800 transfected cells from six independent experiments as in (a) showing conductin condensation. c Western blotting for endogenous (endog.) conductin under native conditions in lysates of SW480 cells treated with indicated GBZ concentrations overnight. d Western blotting for β-catenin, conductin, and α-tubulin (loading control) in hypotonic lysates of WT SW480 and AXIN2/Conductin knockout cells (Cdt KO), which were treated with indicated concentrations of GBZ for 48 h. Numbers below the blots show the relative protein amounts normalized to α-tubulin. e, f Quantification of β-catenin degradation (e) and expression of the β-catenin target gene conductin (f) in WT and knockout cells after GBZ treatment based on five independent experiments as in (d). g mRNA expression of the β-catenin target genes AXIN2 and LGR5 relative to GAPDH in SW480 and DLD1 cells treated with indicated concentrations of GBZ. h–j Luciferase activity (TOP/FOP) after treatment with indicated GBZ concentrations in U2OS cells without and with Wnt3a stimulation (h), in DLD1 cells without and with conductin knockdown (i), and in SW480 WT and AXIN2/Conductin knockout cells (j). Western blots in i show the efficiency of conductin knockdown. To facilitate comparisons of the GBZ effects in cells with (black bars) and without conductin (siCdt/Cdt KO, red bars) in (i, j), the initial luciferase activities without GBZ were set to 100% for both conditions, and the luciferase activities with GBZ treatment are presented relative to the respective initial activity. Results are mean ± SEM (n = 6 [b], n = 5 [e, f], n = 3 [g], n = 4 [h–j]). *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided Student’s t-test). Molecular weight is indicated in kDa (c, d, i). Source data and exact p-values are provided as source data file.

Fig. 7. GBZ treatment inhibits the growth of colorectal cancer cells.

a Intestinal organoids grown within 120 h in the absence (−) or presence of 50 µM GBZ. Scale bar: 200 µm. b Quantification of small (S), medium (M), and large (L) organoids grown in the presence of 0, 10, or 50 µM GBZ in three experiments as in (a). Numbers above the bars indicate the total number of intact organoids. c Colonies grown from SW480, DLD1, and U2OS cells within 96 h in the presence of 0, 10, and 50 µM GBZ, stained by ethidium bromide and visualized with UV light. Scale bar: 0.5 cm. d Automated quantification of colony numbers and sizes from four independent experiments as in (c). e, h MTT color intensity reflects the number of viable SW480 cells growing for 72 and 96 h in the presence of 0 and 50 µM GBZ. WT cells are compared to AXIN2/Conductin knockout (Cdt KO) cells (e), and control siRNA transfected cells (−) to siGαi2 transfected cells (h). Each dot represents a technical replicate (n = 4). f, i Growth reduction by GBZ of four independent experiments as in (e, f) or in (h, i) (growth reduction is calculated as the difference between the MTT color intensities without and with GBZ treatment, and presented as a percentage relative to the MTT intensities without treatment). g Reduction of cell divisions of SW480 WT and AXIN2/Conductin knockout (Cdt KO) cells treated with indicated GBZ concentrations compared to untreated cells, calculated based on four independent CFSE dilution experiments as shown in Supplementary Fig. 15. j MTT color intensity reflects the number of viable SW480 cells expressing similar levels of GFP (−, control) or Gαi2-GFP after cell sorting and growth for 72 and 96 h (see Methods for details). The dots represent independent experiments (n = 4), and the lines connect values of GFP and Gαi2-GFP expressing cells of the same experiment. Results are mean ± SEM (n = 3 [b], n = 4 [d, f, g, i, j]). *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided Student’s t-test). Source data and exact p-values are provided as source data file.

Fig. 8. GBZ treatment inhibits intestinal tumor growth in vivo.

a Volumes of SW480 xenograft tumors growing in BALB/c nude mice that were untreated (n = 8) or treated with 50 µM GBZ in the drinking water (n = 8). b Growth of the individual tumors (dots) calculated as percentage difference of endpoint tumor volume relative to tumor volume before treatment. Note the shrinkage of three GBZ treated tumors (n[untreated] = 8, n[treated] = 8). c Endpoint weight measurement of individual tumors (dots). Dotted lines connect the tumors that were paired according to similar sizes and appearance before the treatment (n[untreated] = 8, n[treated] = 8, see Methods for details). d Excised tumors were ordered to match the pairs of treated (upper row) and untreated tumors (lower row). e–i Analysis of intestinal tumors in untreated (n = 6, black dots) and GBZ treated (n = 7, red dots) APC^Min mice: mean tumor size (e), the relative frequency distribution of tumor sizes (f), the absolute number of tumors exceeding 3 mm² per 10 cm intestine (g), representative images showing large tumors in untreated animals (arrowheads) and small tumors in treated animals (arrows, scale bar: 0.5 cm) (h), and the total number of tumors per 10 cm intestine (i). Results are mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 (two-sided Student’s t-test). Source data and exact p-values are provided as source data file.

Fig. 9. Working model for GBZ inhibiting tumor growth.

Activation of Gαi2 by GBZ via G-protein coupled receptors induces polymerization of diffusely distributed conductin into condensates, thereby promoting conductin-mediated β-catenin degradation and, consequently, decreasing cell proliferation.