SqLC is thought to arise from excessive proliferation of K5+ ABSCs and progress through a series of histologically-defined premalignant intermediates as a product of dysfunctional homeostatic mechanisms (Ishizumi et al., 2010). Further, β-catenin post-translational modifications have been shown to regulate its subcellular localization and signaling in other contexts (Fang et al., 2007; Huber and Weis, 2001; Rhee et al., 2007). To determine whether Wnt/β-catenin signaling is dysregulated during human SqLC progression, we performed immunofluorescence (IF) staining and analysis on human patient airway samples for varying pathway components. In contrast to its nuclear expression in only 5.2% of histologically normal human airway epithelia, β-catenin phosphorylated at Y489 (p-β-catenin^Y489) uniquely localized to the nucleus in 35.7% of human airway premalignant lesions (PMLs) and 50.9% of SqLC patient samples (Figures 1A and 1B). Further, while only 7.2% of basal cells from normal human patient airways contained nuclear p-β-catenin^Y489, 38.3% and 62.7% of ABSCs exhibited nuclear p-β-catenin^Y489 in PMLs and SqLCs, respectively (Figure 1C). These findings stood in contrast to other phosphorylated forms of β-catenin that only localized to the membrane or cytoplasm (Figures S1A–S1C).

Utilizing Porcupine IF staining as an indicator of Wnt secretion by airway cells (Tammela et al., 2017), we additionally observed minimal epithelial and no stromal expression of Porcupine in normal human airways (Figures 1D and 1E). However, Porcupine was expressed in the epithelium of 50% and 100% of PMLs and SqLC samples analyzed, respectively (Figure 1E). We further quantified the percentage of ABSCs that were also Porcupine⁺ in these samples. While 5.8% of ABSCs were Porcupine⁺ in normal human airways, this increased to 34.8% and 70.1% of ABSCs in PML and SqLC patient samples (Figure 1F). We also observed Porcupine staining in the stromal compartment during stepwise progression to SqLC (Figure 1D). While zero of the five normal human airways examined had positive stromal Porcupine staining, 16.7% PMLs and 62.5% SqLCs displayed positive stromal Porcupine expression (Figure 1G). Consistent with these findings, 7.2% of stromal cells in PMLs and 24.3% of stromal cells in SqLCs were also Porcupine⁺ (Figure 1H). These studies collectively elucidate the emergence of a dysregulated Wnt/p-β-catenin^Y489 signaling axis during stepwise progression to human SqLC.

To determine whether the pathobiology observed in human patients can be modeled in vitro, we first isolated and seeded mouse ABSCs (mABSCs) in collagen-coated transwells. On day 1, mABSCs were then treated with the GSK3β inhibitor CHIR99021 (CHIR) under submerged culture conditions until day 4 to activate canonical Wnt signaling (Figure 2A). We performed EdU (5-ethynyl-2′-deoxyuridine) proliferation assays from mABSCs treated with CHIR and observed statistically significant increased ABSC proliferation in comparison to DMSO-treated controls (Figures 2B and 2C).

We next inquired whether there were changes in the ability of ABSCs to differentiate to the ciliated cell fate in vitro following Wnt activation. To this end, mABSCs were isolated and treated with CHIR under submerged culture conditions for 4 days as shown in the schematic presented in Figure 2D. On day 4, media from the apical chambers were removed, and mABSCs were cultured under air-liquid interface (ALI) differentiation conditions with CHIR until day 14 (Figure 2D). Strikingly, mABSCs treated with CHIR exhibited a dose-dependent heaping morphology that resembles the PMLs seen in the airways of patients (Figure 2E). Further, mABSCs treated with two independent GSK3β inhibitors (CHIR and GSK3XV) displayed a significant reduction in the percentage of ciliated cells, indicated by the absence of acetylated β-tubulin, and an increased pool of K5⁺ mABSCs (Figures 2F and 2G). We additionally observed that human ABSCs (hABSCs) treated with CHIR for 21 days (Figure S2A) similarly exhibited abolished differentiation to the ciliated cell fate and an increased pool of ABSCs (Figures S2B and S2C).

We next sought to determine the extent of Wnt/β-catenin signaling activation upon dysregulated ABSC homeostasis by performed IFs on mABSC cultures treated with CHIR. We observed increased nuclear p-β-catenin^Y489 relative to DMSO-treated control cultures (Figure 2H). In contrast, other phosphorylated forms of β-catenin (p-β-catenin^Y654, p-β-catenin^S552, and p-β-catenin^S33,S37,T41) remained primarily cytoplasmic or membranous in the submerged phase of culture (Figures S2D–S2G). Additionally, GSK3β inhibition and recombinant Wnt3a increased TCF/LEF activity measured by a luciferase reporter in comparison to DMSO-treated cultures (Figure 2I). To assess the possibility that differing levels of Wnt signaling could produce phenotypic differences in vitro, mABSCs were treated with varying concentrations of recombinant Wnt3a under ALI conditions. Interestingly, we observed that medium levels of Wnt signaling (100 ng/mL) increased the proportion of ciliated cells, while high levels of Wnt signaling (500 ng/mL) produced fewer ciliated cells at the expense of K5⁺ mABSC expansion (Figures 2J and 2K). Collectively, our results demonstrate that ABSC homeostasis is controlled by perturbations in canonical Wnt signaling, and we have developed a tractable system in which we can monitor the biology underlying ABSC hyperplasia.

In light of our findings that Wnt/β-catenin signaling drives ABSC hyperproliferation, we sought to identify small-molecule inhibitors of the pathway that could restore airway homeostasis. To this end, we conducted a high-throughput drug screen that would decrease Wnt/β-catenin signaling activity and ABSC proliferation (Figure 3A). BEAS2B cells, a normal human bronchial epithelial cell line, were transduced to stably express a TCF/LEF luciferase reporter that could be induced by treatment with CHIR after 24 h (Figure 3B). The Z′ factor for this screen was calculated to be 0.82, indicating a reliable platform to identify target compounds. BEAS2B cells were treated with DMSO, 5 μM CHIR, or 5 μM CHIR + 10 μM screen compound for 24 h, after which TCF/LEF activity was measured. Hoechst staining was also conducted in parallel to assess for toxicity. Compounds that yielded fewer than 80% of nuclei compared to DMSO controls by Hoechst staining were considered toxic and removed from further analysis. We screened 20,000 small molecules and initially identified 70 compounds that decreased TCF/LEF activity, 8 of which were nontoxic, and 4 compounds that exhibited dose dependence. Compounds A–C reproducibly decreased TCF/LEF activity and were nontoxic by CellTiter-Glo measurements (Figures S3A–S3F) but did not modulate ABSC proliferation (data not shown). However, Wnt inhibitor compound 1 (WIC1) was a small molecule that decreased TCF/LEF luciferase reporter activity (Figure 3C) and displayed no toxicity by Hoechst stain (Figure 3D). Further, BEAS2B cells treated with a dose course of WIC1 for 24 h showed a stepwise decrease in TCF/LEF activity (Figure 3E) and remained nontoxic even at 10 μM, as measured by the CellTiter-Glo assay (Figure 3F).

In light of the structure of WIC1 (Figure 3G), we conducted structure-activity relationship (SAR) studies with an additional 160 small molecules that bore structural resemblance to WIC1. We conducted TCF/LEF luciferase dose-course assays with these compounds and identified 8 additional small molecules that reproducibly decreased reporter activity, named WIC2–WIC9 (Figures 3H and S3G). Analysis of the compound hits WIC2–WIC9 from our SAR study indicated a common dihydrocoumarin pharmacophore that was substituted with a carboxy moiety in 3 position that in turn was conjugated to a moiety that had an aniline as maximum common substructure. Together, our results reveal members of a well-known, important pharmacophore for which we have identified a role in modulating canonical Wnt signaling.

As recent studies have identified Wnt inhibitors that mechanistically act at varying levels of the pathway (Fang et al., 2016; Hwang et al., 2016), we measured and compared BEAS2B TCF/LEF luciferase reporter activity following 24 h of treatment with 5 μM CHIR or 5 μM CHIR plus a dose course of WIC1 against other published Wnt inhibitors MSAB, LF3, and ICG001. We observed that while MSAB was the only inhibitor that decreased TCF/LEF activity at 10 nM, WIC1 remained the most potent inhibitor at 100 nM and 1 μM (Figure 3I). Further, all Wnt inhibitors assayed, except WIC1, were toxic at higher concentrations, as measured by CellTiter-Glo (Figure 3J). These data collectively indicate that WIC1 is a more potent Wnt inhibitor across a wider range of concentrations at which other known Wnt inhibitors display cellular toxicities.

To assess whether WIC1 could reverse Wnt/β-catenin-induced primary ABSC hyperproliferation, we treated hABSCs with DMSO, CHIR, or CHIR + WIC1 under submerged conditions for 4 days. Cultures treated with CHIR + WIC1 displayed a significant reduction in ABSC proliferation measured by EdU incorporation (Figures 4A and 4B). We next sought to understand if WIC1 could promote ABSC differentiation to ciliated cells under ALI conditions. Relative to DMSO controls, mABSC cultures treated with 100 nM WIC1 significantly induced ciliated cell differentiation (Figures 4C and 4D). Interestingly, however, higher concentrations of WIC1 (1 μM) decreased the percentage of ciliated cells in culture (Figures 4C and 4D), consistent with our observations that tightly regulated levels of Wnt signaling are critical for proper regulation of airway homeostasis (Figures 2J and 2K).

We next inquired how the effects of WIC1 on ABSC proliferation and differentiation compared to the Wnt inhibitors MSAB, LF3, and ICG001. To this end, we treated mABSCs under submerged culture conditions for 4 days with DMSO, 1 μM CHIR, and 1 μM CHIR + 1 μM WIC1, 1 μM MSAB, 1 μM LF3, or 1 μM ICG001. We observed that cultures treated with CHIR and all of the Wnt inhibitors, including WIC1, decreased mABSC proliferation to a similar degree, as measured by EdU incorporation (Figures S4A and S4B). These data indicate that WIC1 regulates ABSC proliferation in a manner similar to other known Wnt signaling inhibitors. Additionally, we performed ALI differentiation assays with mABSCs treated with DMSO and two different concentrations (100 nM and 1 μM) of WIC1, MSAB, LF3, or ICG001. These studies revealed that 100nM of WIC1, LF3, and ICG001 induced a greater proportion of ciliated cells under ALI conditions in comparison to DMSO-treated cultures (Figures S4C and S4D). However, treatment with 1 μM WIC1, LF3, or ICG001 had no effect on the percentage of ciliated cells in culture (Figures S4C and S4D). Interestingly, neither 100 nM nor 1 μM MSAB increased differentiation to the ciliated cell fate (Figures S4C and S4D). Taken together, our findings illustrate that low-dose WIC1 acts in a similar fashion to a subset of other known Wnt inhibitors in promoting ciliogenesis.

To develop an understanding of the mechanism by which WIC1 modulates Wnt signaling, we first performed qRT-PCR to assess target gene expression in the context of BEAS2B cells treated with DMSO, 5 μM CHIR, or 5 μM CHIR + 1 μM WIC1. While we observed a notable increase in mRNA expression of known downstream Wnt signaling target genes CCND1, MYC, and CTNNB1 with CHIR treatment, addition of WIC1 to these cultures resulted in a significant decrease in their expression (Figure 4E). Moreover, recent work by Haas et al. demonstrated that high levels of Wnt/β-catenin signaling activate ΔN-TP63 expression to promote basal cell proliferation (Haas et al., 2019). As such, we first sought to understand whether treatment with WIC1 under settings of high Wnt signaling could modulate expression of TP63. qRT-PCR studies from RNA isolated from BEAS2B cells identified an increase in TP63 expression with the addition of CHIR that was reduced upon addition of WIC1 (Figure S4E). Further, RNA was also isolated from primary mABSC cultures on day 9 of ALI, and we found induction of Trp63 expression with CHIR treatment that was substantially reversed with WIC1 treatment (Figure 4F). To further dissect the mechanism by which WIC1 acts, we next assessed its effect on phosphorylation of β-catenin at Y489. To this end, we treated BEAS2B cells for 24 h with DMSO, 5 μM CHIR, or 5 μM CHIR + 1 μM WIC1. Strikingly, while CHIR treatment induced nuclear localization of p- β-catenin^Y489, addition of WIC1 to cultures abolished its nuclear localization (Figures 4G and 4H). Taken together, these data offer insight that WIC1 modulates its phenotypic effects on airway homeostasis, in part, via TP63/p-β-catenin^Y489-dependent mechanisms.

Further information and requests for resources and reagents should be directed to and will be fulfilled by lead contact, Brigitte Gomperts (bgomperts@mednet.ucla.edu). This study did not generate unique reagents.

Statistical methods relevant to each Figure are outlined in the Figure Legend. Statistical anlaysis was performed using GraphPad Prism software version 7–8 (GraphPad, San Diego, CA, USA). Multiple high-power fields were imaged to increase the number (n) of samples used for quantification purposes. Data are presented as the mean ± standard error of the mean (SEM) values. Student’s t tests were used to assess statistical significance. Calculated p values are indicated on individual figures.

This study did not generate any unique datasets or code.