Sema3C signaling is an alternative activator of the canonical WNT pathway in glioblastoma – Nature Communications

Combined inhibition of Wnt and Sema3C pathways improves survival

Previous studies in preclinical mouse models of GBM suggest minimal if any survival benefit with Wnt inhibition alone despite overwhelming data of improved tumor control in vitro^8,13. The Porcupine inhibitor LGK974 blocks Wnt secretion and is under clinical investigation^33,34,35. Since GSCs are more resistant to therapeutic agents^36,37, we first tested the sensitivity of GSCs to LGK974. We found that the half maximal effective dose (EC50) of LGK974 is about 6.5 µM in GSCs (Supplementary Fig. 1a). Treatment of GSCs with LGK974 reduced GSC self-renewal and viability in a dose-dependent manner, demonstrating the sensitivity of GSCs to this specific Wnt pathway inhibitor (Supplementary Fig. 1b).

Considering that many drugs are unable to pass through the blood-brain-barrier and blood-tumor-barrier, we first asked whether LGK974 can engage its target in a GSC-derived orthotopic mouse model of GBM. After confirming tumor growth by bioluminescence imaging, mice were randomized to receive LGK974 or placebo control by oral gavage at two different dose levels (Supplementary Fig. 1c). The higher dose of 10 mg/kg twice a day induced diarrhea in all of the animals after 10 days of treatment, consistent with known intestinal toxicity of Wnt inhibitors³⁸, and led to premature death. Mice tolerated the lower dose (5 mg/kg twice a day) without clinically significant toxicity. We expanded our studies to larger numbers of mice (Fig. 1a). Analysis of the tumors from treated animals showed a significant reduction in the expression of the Wnt target gene TCF1³⁹ (Fig. 1b). These data suggest that at a well-tolerated dose, LGK974 can indeed pass the blood-brain-barrier and blood-tumor-barrier to engage its target. However, we did not see any survival benefit (Fig. 1a, p = 0.4973) even though TCF1 expression was reduced (Fig. 1b) in tumors of treated animals. Our data suggest that blocking Wnt secretion alone is insufficient to reduce tumor growth in vivo and that other pathways contribute to resistance.

Cancer stem cells have evolved mechanisms to reactivate developmental pathways to ensure their survival and maintenance⁴⁰. We next assessed whether other signaling pathways that are often deregulated in GBM, in particular in GSCs, can act as a driver of canonical WNT pathway. Since Sema3C is overexpressed in the overwhelming majority (>85%) of GBM and Sema3C signaling is frequently activated in GSCs²⁵, we assessed the possibility that Sema3C could regulate canonical Wnt signaling.

To this end, we first investigated the expression pattern of TCF1, the Wnt target gene and β-catenin binding partner, and Sema3C in human GBM samples (Fig. 1c and Supplementary Fig. 1d, e). By dual immunohistochemical staining, we found that the expression of both proteins is highly correlated (Fig. 1c, d; Pearson correlation coefficient, r = 0.82, R² = 0.66, p < 0.0001), suggesting that they may function in the same signaling axis.

To determine potential interactions between Sema3C and Wnt signaling, we used a genetic approach. We established tumors in which Sema3C, TCF1, or both Sema3C and TCF1 were knocked down. In two different models, knockdown of either Sema3C or TCF1 alone reduced tumor growth and extended animal survival compared to control animals (Fig. 1e–g, Supplementary Fig. 1f, g). Silencing both Sema3C and TCF1 resulted in the greatest tumor control and longest animal survival. Ki67 staining was markedly reduced in tumors in which Sema3C and TCF1 were silenced, supporting a reduction in tumor cell proliferation (Fig. 1f). Together, these data support that inhibition of both Sema3C and Wnt pathways synergize to improve survival more than single pathway inhibition alone.

Sema3C is indispensable for the self-renewal of GSCs

Since the Wnt pathway facilitates stem cell self-renewal, and our study suggests a link between Sema3C and Wnt pathways, we investigated the functional role of Sema3C in GSC self-renewal. We assessed tumorsphere formation and performed in vitro extreme limiting dilution assays using GSC lines in which Sema3C was silenced. Sema3C knockdown (shSema3C) reduced tumorsphere formation (Fig. 2a, b) and reduced self-renewal capacity compared to control non-targeting knockdown (shNT) (Fig. 2c, d). The frequency of GSCs was reduced 6- to 13-fold after silencing Sema3C, depending on the cell line (Fig. 2c, d). In addition, Sema3C knockdown significantly reduced proliferation, as evidenced by reduced incorporation of the thymidine analog ethynyldeoxyuridine (EdU) into DNA (Fig. 2e, f). This reduction in proliferation contributed to a decline in cell number after Sema3C knockdown (Fig. 2g, h). These data suggest that Sema3C plays an essential role in GSC self-renewal.

a, b Left panels: representative tumorsphere images of 387 (a) and 4121 (b) GSCs after control shNT, shSema3C#1, and shSema3C#2 knockdown. Scale bar 100 μM. Right panels: quantification of tumorspheres after Sema3C knockdown (n = 4 biological replicates in each group, compared with control unadjusted p < 0.0001, error bars: S.D. Scale bar 100 μM). c, d In vitro extreme limiting dilution assay in Sema3C knockdown 387 GSCs (c) and 4121 GSCs (d). Tables show estimated stem cell frequencies in control shNT, shSema3C#1, and shSema3C#2 knockdown GSCs with 95% confidence intervals (n = 24 replicates in each dose, compared with control unadjusted p < 0.0001, error bars: S.D.). e, f Left panels: representative immunofluorescence images of EdU incorporation assay in 387 (e) and 4121 (f) GSCs after control shNT, shSema3C#1, and shSema3C#2 knockdown. Right panels: quantification of EdU⁺ cells over total cells (EdU in red, DAPI in blue, n = 3 independent experiments, 387 GSCs control vs. shSema3C#1 unadjusted p = 0.0002; control vs. shSema3C#2 unadjusted p = 0.0006. 4121 GSCs control vs. shSema3C#1 unadjusted p < 0.0001; control vs. shSema3C#2 unadjusted p = 0.0053, error bars: S.D. Scale bar 50 μM). g, h Cell viability of 387 and 4121 GSCs after Sema3C knockdown (n = 6 biological replicates for each group, compared with control unadjusted p < 0.0001, error bars: S.D.). Source data are provided as a Source data file.

Sema3C signaling regulates the Wnt pathway

We next tested the effects of Sema3C on canonical Wnt signaling. Expression of key Wnt target genes including TCF1 and c-Myc were down-regulated at the mRNA and protein levels when Sema3C was silenced (Fig. 3a, b, Supplementary Fig. 2a). We then overexpressed FLAG-tagged full-length Sema3C in GSCs. Overexpression of Sema3C led to increased expression of multiple Wnt target genes including Axin2, CCND1, c-Jun, C-myc, and TCF1 at the mRNA level (Fig. 3c). Changes in expression of a subset of these targets were confirmed at the protein level (Supplementary Fig. 2b). Together, these data suggest that Sema3C can regulate the canonical Wnt pathway.

a Western blots of TCF1, c-Myc, and c-Met proteins after Sema3C knockdown. Western blots were repeated at least twice. b qRT-PCR of TCF1 and C-myc in 3359 and 4121 GSCs after Sema3C knockdown (n = 4 independent experiments, Mann Whitney U-test p < 0.03, error bars: S.D.). c qRT-PCR of Axin2 (3359 GSCs p < 0.0001; 4121 GSCs p = 0.013), CCND1 (3359 GSCs p = 0.0378; 4121 GSCs p = 0.045), c-Jun (3359 GSCs p = 0.0004; 4121 GSCs p = 0.035), C-myc (3359 GSCs p = 0.04; 4121 GSCs p = 0.0063) and TCF1(3359 GSCs p = 0.019; 4121 GSCs p = 0.0059) in GSCs expressing FLAG vector or FLAG-Sema3C (n = 3 independent experiments, Two-tailed T-test, error bars: S.D.). Source data are provided as a Source data file.

Sema3C regulates β-catenin nuclear translocation

In canonical Wnt signaling, translocation of β-catenin into the nucleus is a critical event that leads to transactivation of Wnt target genes¹. We therefore investigated the role of Sema3C in regulating β-catenin nuclear localization. We performed cellular fractionation assays to separate cytosolic and nuclear fractions of GSCs, followed by immunoblotting of β-catenin. The nuclear fraction of β-catenin was significantly reduced when Sema3C was silenced in four different GSC lines (Fig. 4a, Supplementary Fig. 3). We next complemented these fractionation studies with immunofluorescence imaging of β-catenin. Knockdown of Sema3C in GSCs reduced the fraction of cells exhibiting nuclear β-catenin by at least 50% (Fig. 4c, d). Conversely, overexpression of Sema3C led to an increase in cells with β-catenin nuclear localization (Fig. 4b). These data strongly suggest that Sema3C regulates the subcellular localization of β-catenin.

a Western blots of β-catenin in cytosolic and nuclear fractions (left) of GSCs after Sema3C knockdown (right). Western blots were repeated at least twice. b Quantification of nuclear β-catenin positive cells of GSCs expressing FLAG vector or FLAG-Sema3C (n = 3 independent experiments in 387 GSCs Two-tailed T-test p = 0.0066, error bars: S.D.; n = 4 independent experiments in 4121 GSCs, Two-tailed T-test p = 0.0088, error bars: S.D.). c, d Left panels: representative immunofluorescence images of β-catenin (green) in 387 (c) and 4121 (d) GSCs after control shNT and shSema3C knockdown. Nuclei were counterstained with DAPI (blue). Right panels: quantification of nuclear β-catenin positive cell fraction (n = 3 independent experiments, Two-tailed T-test, p = 0.0012 (387 GSCs); p < 0.0001 (4121 GSCs) error bars: S.D. Scale bar 50 μM). Source data are provided as a Source data file.

Activation of Rac1 by Sema3C facilitates β-Catenin nuclear translocation

Sema3C binds to the NRP1/PlexinD1 receptor complex in GSCs²⁵. The PlexinD1 co-receptor functions in signal transduction and is essential in GSC maintenance. We reasoned that silencing PlexinD1 would phenocopy the effects of Sema3C knockdown on Wnt signaling. Indeed, we observed a reduction in TCF1 and c-Myc expression in three independent GSCs in which PlexinD1 was silenced (Fig. 5a).

a Western blots of TCF1 and c-Myc proteins after PlexinD1 knockdown in GSCs. b Western blots of Rac1-GTP and Rac1 after Sema3C knockdown in GSCs. c Western blots of TCF1 and c-Myc in NSC23766-treated GSCs. d Western blots of TCF1 after Rac1 knockdown in GSCs. e Western blots of TCF1 and c-Myc in GSCs expressing FLAG-Rac1-Q61L in the setting of Sema3C knockdown. Immunoblots using whole cell lysates (left) and cytosol and nuclear fractions (right) are shown. Source data are provided as a Source data file. Western blots were repeated at least twice.

We previously demonstrated that Rac1 is activated upon Sema3C binding to its receptor complex, and Rac1 is an essential mediator of Sema3C-dependent glioblastoma progression²⁵. The small GTPase Rac1 can regulate the Wnt pathway by facilitating β-catenin nuclear translocation³². Considering the effects of Sema3C on GSC self-renewal, we next assessed the potential convergence of Sema3C and canonical Wnt signaling in GSCs as mediated through Rac1. We first silenced Sema3C and assessed active Rac1 by Rac1-GTP pull-down assays. Rac1-GTP was significantly reduced in GSCs in which Sema3C was silenced (Fig. 5b), consistent with our previous findings²⁵.

We next tested the ability of Rac1 to regulate Wnt signaling. GSCs were treated with increasing concentrations of the Rac1 inhibitor NSC23766⁴¹. After inhibition of Rac1, TCF1 and c-Myc expression were reduced in a dose-dependent manner, phenocopying Sema3C knockdown (Fig. 5c). We observed a similar reduction of TCF1 protein expression when we silenced Rac1 in GSCs (Fig. 5d). These studies suggest that Rac1 mediates Sema3C-induced Wnt signaling.

We next tested the ability of constitutively active Rac1 (Rac1-Q61L)⁴² to restore the expression of TCF1 and c-Myc, and nuclear localization of β-catenin in the setting of Sema3C knockdown. As expected, active Rac1 rescued TCF1 and c-Myc expression (Fig. 5e, left) and increased nuclear β-catenin in all three GSC lines tested (Fig. 5e, right). These data support that Sema3C can regulate the canonical Wnt pathway through activation of Rac1.

High Sema3C expression drives canonical Wnt signaling despite inhibition of Wnt ligand secretion

Our data suggest that Sema3C can drive canonical Wnt signaling by facilitating ß-catenin nuclear translocation irrespective of Wnt ligand-receptor interaction. To test this possibility, we performed Sema3C overexpression studies in the presence of a Wnt pathway inhibitor. Treatment of GSCs with the Porcupine inhibitor, LGK974, reduced expression of TCF1 and c-Myc expression in a dose-dependent manner (Fig. 6a). We next overexpressed FLAG-tagged Sema3C in GSCs followed by treatment with LGK974. Sema3C overexpression rescued LGK974-mediated reduction in TCF1 expression (Fig. 6b). Together, these data support that high levels of Sema3C signaling can drive canonical Wnt signaling to bypass Wnt ligand inhibition.

a Western blots of TCF1 and c-Myc in GSCs treated with Porcupine inhibitor LGK974. Cells were treated with LGK974 or vehicle control for 24 h. b Western blots of TCF1 in LGK974 treated GSCs expressing FLAG vector or FLAG-Sema3C. Cells were treated with LGK974 100 μM or vehicle control for 24 h. Source data are provided as a Source data file. Western blots were repeated at least twice.

Combined Sema3C and Wnt pathway inhibition improve GBM control

GSCs are dependent on the Sema3C pathway and Wnt signaling for their maintenance. We reasoned that inhibition of both Sema3C and Wnt pathways would achieve better GSC control than inhibition of either pathway alone. To this end, we silenced Sema3C and TCF individually or together. Silencing both Sema3C and TCF1 led to the greatest reduction in GSC self-renewal capacity compared to single knockdown in limiting dilution assays, suggesting synergy with dual inhibition (Fig. 7a). Similarly, combined Sema3C knockdown with the Wnt pathway inhibitor LGK974 disrupted Wnt signaling more than single pathway treatment alone (Fig. 7b). These in vitro data support our animal studies that dual inhibition of Sema3C and Wnt pathways reduces GSC self-renewal and GSC-mediated tumorigenesis (Fig. 1g, Supplementary Fig. 1f).

a In vitro extreme limiting dilution assay in shSema3C, shTCF1 or both knockdowns. Tables below show estimated stem cell frequencies in control shNT, shSema3C, shTCF1, and shSema3C + shTCF1 knockdown GSCs (n = 8 technical replicates in each dose, at least three biological replicates, ELDA test, comparing to shSema3C + shTCF1 knockdown GSCs, 387 GSCs: shNT p < 0.0001; shSema3C P = 0.0003; shTCF1 p = 0.0002; 4121 GSCs: shNT p < 0.0001; shSema3C p < 0.0001; shTCF1 p = 0.028). b Western blots of TCF1 and c-Myc in shSema3C knockdown GSCs treated with LGK974 100 μM or vehicle control for 24 h. Western blots were repeated at least twice. c Model of Sema3C regulation of the Wnt pathway in GSCs. Despite upstream Wnt pathway inhibition, Sema3C binds to the Neuropilin 1 (NRP1)—PlexinD1 receptor complex pathway to activate Rac1. Active Rac1 (Rac1-GTP) facilitates β-catenin nuclear translocation to drive Wnt target gene transcription. Source data are provided as a Source data file.

Based on our findings, we propose a model by which Sema3C can drive Wnt signaling irrespective of Wnt ligand binding to its receptor (Fig. 7c). Sema3C signaling activates Rac1 to facilitate β-catenin translocation into the nucleus to initiate downstream Wnt target gene expression. Therefore, GSCs can bypass Wnt ligand-receptor interaction and instead directly engage downstream Wnt pathway effectors to promote their self-renewal and facilitate tumor growth.