(A) Schematic of the screening assay using the BAR reporter. (B) Establishment of the optimal dose of BIO for the enhancer screen using RKO-BAR cells. RKO-BAR cells were mock infected (No LV-shRNA), or were infected with lentiviral vectors that expressed no shRNA (LV-control) or shRNAs targeting either GSK3β or Axin1. Forty eight hours after transduction, cells were exposed to a titration of BIO (GSK3i), and luciferase activity was measured after 24 hr of exposure to drug. Silencing of GSK3β or Axin1 enhanced luciferase activity at doses of BIO that showed no detectable luciferase activity in mock and control vector infected cells (green box). This optimized dose range of BIO (300–500 nM) was used in our screen and follow up studies. (C) Volcano plot of the screen using a lentiviral shRNA library targeting the druggable genome. Samples were screened in triplicate in the presence of 312 nM BIO. shRNA vectors that increased luciferase activity by 3-fold or greater with a P value less than 0.01 are highlighted as top hits. Among this set of hits, known members of the β-catenin destruction complex (including GSK3β) are colored and labeled in orange, whereas potentially novel GSK3 modulators including the two DHFR shRNAs are colored maroon. The two PSMD7 shRNAs are colored maroon with an orange outline. An expanded hit list with full statistical analysis is provided in Table S1.

(A) The experiment was performed as described for BIO in figure legend 2 with the substitution of an alternative GSK3 inhibitor. (B) RKO-BAR cells were exposed to multiple activators of β-catenin signaling (GSK3 specific inhibitors BIO and GSK3i VIII, lithium chloride (LiCL), and Wnt3A conditioned medium (3ACM)) or controls (DMSO, NaCL, and control conditioned medium (LCM)). Protein lysates were harvested and subjected to Western Blot analysis with a GSK3β phospho-serine 9 specific antibody. The enhancement of BAR activity by DHFR inhibition correlated with agents that induced phosphorylation of GSK3 at inhibitory residue serine 9. The result is representative of two separate experiments.

RKO-BAR cells were treated with the proteasome inhibitor MG132 alone (1 µM), or in combination with other compounds as depicted, and subjected to immunoprecipitation with a β-catenin specific antibody followed by Western blot for total and phospho- (Ser33/37/Thr41) β-catenin. Exposure of cells to suboptimal BIO (312 nM) plus methotrexate (312 nM) led to a greater decrease in phosphorylated β-catenin than exposure to the same dose of BIO alone (compare lanes 4 and 5). As observed with β-catenin accumulation (Figure 2), further exposure to adenosine (100 µM) neutralized the impact of methotrexate on BIO-mediated inhibition of β-catenin phosphorylation. The dashed line denotes that the adenosine treated sample is from a separate part of the same blot. This result is representative of three separate experiments.

Cytokine production in PBMCs preincubated for 1 hr in medium only (No Drug), 5 nM BIO (GSK3i), 10 nM methotrexate (MTX), or the combination of the two compounds (BIO + MTX) and then stimulated with 300 ng/ml LPS for 20 hr. No LPS controls are shown on the far right of each graph for reference. Cell free supernatants were analyzed by ELISA for production of TNFα (A), IL-6 (B), IL-12 (C), or IL-10 (D). *, P<0.05, **, P<0.01, and ***, P<0.001, compared with LPS alone. Results represent the mean±SD of 6 replicate wells for each sample. The results shown are representative of three separate experiments.

(A) Hit confirmation by the BAR assay. Fresh virus generated for DHFR shRNAs, GSK3β shRNAs, and empty vector controls was used to transduce RKO-BAR cells. 48 hours following transduction cells were exposed to a sub-optimal (non-activating) dose of GSK3i (312 nM) for another 24 hours before measuring pathway activation by luciferase activity. Importantly, all three vectors targeting DHFR (red bars) induced similar levels of activation as shRNAs targeting GSK3β. (B) Examination of the effect of the DHFR inhibitor methotrexate on BAR activation by BIO. RKO-BAR cells were exposed to a titration of methotrexate in the presence (red) or absence (blue) of a non-activating dose of BIO and luciferase activity was measured. In confirmation of the enhancement of BIO activity in cells transduced with shRNA targeting DHFR, inhibition by methotrexate is synergistic with GSK3 inhibition for activation BAR.

(A) RKO-BAR cells were transduced with a lentiviral vector control lacking an shRNA (U6-TERM), with vectors encoding shRNAs targeting β-catenin or GSK3β, or with vectors encoding three distinct shRNAs targeting DHFR and then exposed to BIO (right lanes) or vehicle control (left lanes) 48 hours following transduction. Protein lysates were harvested and subjected to Western Blot analysis with a β-catenin specific antibody. Whereas none of the vectors had an effect in the absence of drug, silencing of GSK3β or DHFR synergized with BIO to induce accumulation of β-catenin (CTNNB1, blot is representative of three separate experiments). (B) RKO-BAR cells were exposed to methotrexate (MTX), BIO (312 nM), or the combination of the two drugs and stained with the nuclear dye Hoechst and with an antibody recognizing β-catenin. Images were collected on an InCELL 1000 high content imager. A synergistic increase in β-catenin expression and nuclear localization was observed upon dual inhibition of GSK3 and DHFR. (C) RKO-BAR cells were exposed to a titration of methotrexate in the presence of 312 nM BIO and stained for β-catenin. β-catenin expression was quantified on an Acumen Explorer laser scanning microplate reader. As shown in the graph, a methotrexate dose-dependent increase in β-catenin levels was observed (red curve, each point represents the average of six replicate samples). The effect of a BAR-activating dose of BIO alone (1 µM) is provided as a reference (blue). (D) RKO-BAR cells were subjected to DHFR, GSK3 (BIO 312 nM), or dual inhibition in the absence or presence of products of the folate metabolism pathway. β-catenin accumulation was analyzed as in panel C. Data represent the mean and standard deviation (SD) of 8 replicate wells for each sample. Addition of purines (adenosine, Ad., or hypoxanthine, Hypo.) or s-adenosylmethionine (SAM), but not thymidine (Thy.) blocked the synergistic effect of dual inhibition. (E) RKO-BAR cells were subjected to DNMT inhibition (DNMTi) with 5′ Aza-2′-deoxycytidine (750 nM), BIO (312 nM), or both inhibitors in combination. Luciferase activity was measured 16 h following drug addition. Data represent the mean±SD of 4 replicate wells for each sample.

RKO-BAR cells treated with either suboptimal BIO (312 nM) plus methotrexate (red bars), or with an active concentration of BIO (1.5 µM, blue bars), were exposed to a titration of Wortmannin, and luciferase activity was measured 24 h following exposure to drug. Consistent with a requirement for PI3K signaling to mediate the synergistic interaction between DHFR and GSK3 inhibitors, wortmannin significantly reduced BAR activity induced by the combination of the two inhibitors in a dose dependent manner. In contrast wortmannin had modest effect on BAR activity induced by the active dose of BIO alone. Results represent the mean±SD of three replicate wells for each sample.