Foxc1 induces expression of Arhgap36 and activates Hedgehog signalling.

(A) Volcano plot depicting in green differentially expressed genes with Foxc1 expression in NIH3T3 cells. (B) Confirmation of the robust Foxc1-induced increase in Arhgap36 mRNA in NIH3T3 cells by qPCR. (C) Foxc1 drives endogenous Arhgap36 protein expression, and that of Gli1, in NIH3T3-Gli2-mGFP cells to comparable levels of Myc-FLAG Arhgap36 ectopically expressed in parental NIH3T3 cells. (D) Immunofluorescence imaging demonstrates strong endogenous Arhgap36 expression, with the membrane staining in Foxc1-expressing cells recapitulating that of ectopically expressed Arhgap36-MF protein. [RNA-seq: n=3; quantitative Western blots: n=4 replicates; MF denotes Myc-FLAG tagged Arhgap36]

ChIP-seq identification of Foxc1-binding sites at the Arhgap36 locus.

(A) ChIP sequencing with independent anti-Foxc1 antibodies revealed substantial peak overlap for both ChIP samples, consistent with high antibody specificity. Peak calling identified five significant ChIP signal regions within ± 100 kb of Arhgap36 [2 distal, 3 proximal; q-value ≤ 0.05]. (B) Within these Foxc1 ChIP peaks, the discovery algorithm STREME identified two major groups of significantly enriched motifs [p = 2.8 × 10−20, 2.5 × 10−65]. The group 1 position weight matrices are highly similar to known Foxc1 motifs, while the group 2 PWMs very closely resemble the heptanucleotide recognition sequence bound by Fos-Jun transcription factor dimers [most prominently Fosl2]. The distribution of both PWM groups in the vicinity of Arhgap36 is shown on plot A.

Foxc1-induced Arhgap36 reduces levels of protein kinase A catalytic subunit (PKAC).

(A) Foxc1 expression in Gli2-mGFP NIH3T3 cells strongly reduces PKAC, and catalytically active pT197 PKAC, to comparable levels to those observed with ectopic expression of Arhgap36-MF. Quantification shows > 2-fold reduction of PKAC/pT197 PKAC protein levels [Western blots: n=4 replicates]. (B) Immunofluorescent staining demonstrates equivalent reductions in PKAC/pT197 PKAC signal in the cytoplasm and at the basal bodies of cells expressing either Foxc1 or ectopic Arhgap36-MF. [Dashed box: 3x insets, basal body: white arrows].

CRISPRi at potential Foxc1-binding sites diminishes Arhgap36 expression and Hedgehog activity.

(A) In independent CRISPRi-competent clonal cell lines, pools of guide RNAs targeting the Prox-3 and Dist-2 regions, robustly reduced Arhgap36 and Gli1mRNA expression. (B) The reduced Arhgap36 and Gli1 protein expression with CRISPRi sgRNA targeting, implicates the same two ChIP-seq identified peaks in Foxc1’s transcriptional control of the Arhgap36 locus.

Foxc1-induced Hh signalling has reduced dependence on Smoothened.

(A) Elevated levels of Gli1 mRNA in Foxc1-expressing NIH3T3 cells are resistant to inhibition by the Smoothened antagonists sonidegib and cyclopamine. Wild-type NIH3T3 cells stimulated with Smoothened agonist (SAG) and treated with sonidegib, provide a control for inhibitor efficiency. (B) Resistance to Smoothened inhibition is supported by the elevated Gli1 protein levels in Foxc1-expressing Gli2-mGFP NIH3T3 cells treated with sonidegib. Note that expression of Foxc1 induces comparable Gli1 protein levels to vector control cells [pLXSH] treated with SAG; and that levels of Arhgap36 protein itself are unaffected by either sonidegib, or SAG treatment. [qPCRs, quantitative Western blots: n=4 replicates].

Foxc1 promotes ciliary accumulation and decreases phosphorylation of Sufu.

(A) Representative immunofluorescence images demonstrate increased Sufu accumulation at axonemal tips of Foxc1-expressing cells, and NIH3T3 cells that express Myc-FLAG Arhgap36. Note, SAG stimulation per se does not substantially affect ciliary accumulation of Sufu. (B) Distribution of Sufu intensity in individual cilia (C) Mean ciliary Sufu intensity values. These demonstrate that the ciliary Sufu signal in cells expressing Foxc1, and separately Arhgap36, is substantially increased relative to empty vector controls [pLXSH and pLXSN; n = 9 combined experiments]. (D) Decreased phosphorylation of Sufu at the S342 residue in Gli2-mGFP NIH3T3s expressing Foxc1, and NIH3T3 cells expressing Myc-FLAG Arhgap36, relative to vector controls. Expression of Foxc1 also significantly impacts the total protein levels of Sufu, in contrast to the non-significant effect of Arhgap36 [quantitative Western blots: n=4 replicates].

Arhgap36 is required for Foxc1-induced activation of Hh signalling.

(A) Immunofluorescence images demonstrate that two Arhgap36-targeting shRNAs each substantially reduce axonemal tip accumulation of Gli2 in cells that express Foxc1. (B, C) Individual and mean ciliary Gli2 intensity values demonstrate the decrease of ciliary Gli2 signal with Arhgap36 shRNA inhibition [n = 3 combined experiments]. (D) qPCR analyses demonstrate a strong decrease in Gli1 mRNA levels in Foxc1-expressing cells treated with Arhgap36 shRNAs, relative to pLKO.5 control. Note, concordant Gli1 and Arhgap36 expression levels across all conditions.

Overall survival of neuroblastoma patients stratified by Arhgap36 expression levels.

(A-C) Kaplan-Meier plots show poor overall survival with low Arhgap36 expression across three independent neuroblastoma datasets. The significant reduced 5-year survival for cases with low levels of Arhgap36 expression is evident from the range of Hazard Ratios (2.7 – 8.0). Panel (D) shows the same analysis once the data are merged into a single dataset comprising n=1348 individuals (HR 2.8 – 4.8). [Cases stratified into terciles of Arhgap36 expression as shown in boxplots: first tercile (T1) “low”; second (T2) “medium”; third (T3) “high”].

Foxc1 induces expression of Arhgap36 and activates Hedgehog signalling.

(A,B) Stable overexpression of Foxc1 in C2C12 and ATDC5 cells induces strong increases in Arhgap36, and Gli1, mRNA levels.

Foxc1-driven Arhgap36 reduces levels of protein kinase A catalytic subunit (PKAC) in Gli2-mGFP NIH3T3 cells and 3T3-L1 preadipocytes.

(A) Immunofluorescent staining confirms substantial reduction of PKAC signal in Foxc1-expressing Gli2-mGFP NIH3T3 cells. Note correlation between positive Arhgap36 staining and PKAC loss in individual cells. (B) Immunofluorescent staining of Foxc1 and PKAC under the same conditions. Note nearly complete loss of PKAC signal in cells with high Foxc1 nuclear signal. (C) Foxc1 induced expression of endogenous Arhgap36 in 3T3-L1 cells is accompanied by 40-50% reduction in PKAC/pT197 PKAC protein levels. (D) Arhgap36-MycFLAG expression in 3T3-L1 cells strongly reduces PKAC, and catalytically active pT197 PKAC.

The ARHGAP36 locus is located in a region of predominantly closed chromatin.

(A) DNAse I hypersensistivity data from the UCSC genome browser for the 100 kb regions surrounding human ARHGAP36, and GLI1 and PRKACA (for comparison), from diverse human cell lines (ENCODE). DNAse I HS is a marker for open chromatin that is commonly found at active cis-regulatory sequences including promoters and enhancers. At the GLI1 and PRKACA loci, note the multiple, relatively uniform density signal peaks (appearing as dark vertical lines) for the majority of cell lines particularly immediately adjacent to transcription start sites (TSSs), illustrating “open chromatin” that is consistent with expression of both genes in a wide range of tissues. In contrast, strong DNAse I HS signal at the ARHGAP36 locus is observed in the subset of the 75 cell lines of developmental origin marked with a purple vertical bar [Embryonic stem cells (H1hESC, H7hESC) and induced pluripotent stem (iPS CWRU1, iPS NIHi11, iPS NIHi7) cell lines]. The signal density for these is highest adjacent to several ARHGAP36 TSSs, corresponding to the multiple known transcripts, with weak signal levels observed for most of the other cell lines.

ChIP-qPCR validation of Foxc1 ChIP-Seq peaks in the vicinity of Arhgap36 gene.

(A,B) Plots demonstrate read distribution for Foxc1 ChIP-Seq data from Figure 2 [A, entire locus (within ± 100 kb of Arhgap36 ORFs); B, individual peaks (5 kb regions); location of amplicons (a-j) that were analysed by ChIP-qPCR is indicated in red]. (C) ChIP-qPCR data confirm substantial enrichment of signal in all proximal and distal peaks identified by ChIP-Seq (N = 1).

De novo motifs detected in Foxc1 ChIP-Seq peaks using STREME closely resemble known binding motifs of Foxc1 and Fosl2.

(A) De novo motif analysis conducted using STREME tool identified two major groups of significantly enriched motifs (PWMs groups 1 and 2). Group 1 PWMs (ranked top #1 for intersected peaks) share high degree of similarity with known Foxc1 motifs, while Group 2 PWMs (ranked top #2 for intersected peaks) are highly similar to motifs of Fos-Jun family TFs, most prominently Fosl2. Notably, when analysed separately, ChIP-Seq data obtained using each of two anti-Foxc1 antibodies produce highly similar de novo PWMs. Known motifs of Foxc1 and Fosl2 transcription factors were retrieved from JASPAR and HOCOMOCO databases for comparison.

CRISPRi analysis of potential Foxc1-binding sites in the vicinity of Arhgap36 gene.

(A) Plot shows read distribution for the ChIP samples and input chromatin control with five identified peaks (Prox-1 to Prox-3, Dist-1 and Dist-2). Dashed line demonstrates potential relationship between Prox-3 and Dist-2, as promoter and enhancer regions for murine Arhgap36. Panel (B) shows 1) ChIP-Seq coverage at both affected peaks; 2) nearby location of TSS for a major Arhgap36 isoform within the wide Prox-3 peak (approximately 80 – 110 bp downstream of narrow peak Prox-3a, and 500 bp upstream narrow peak Prox-3b); 3) location of individual sgRNA sites used for CRISPRi targeting.

Conservation in Prox 3 and Dist 2 regions of Arhgap36 locus

(A,B) Representation of portions of the Arhagp36 locus corresponding to ChIP-seq peaks Prox-3 and Dist-2. Depicted in order are: ChIP-seq tracks with read coverage for two Foxc1 antibodies, UCSC-derived placental and non-placental mammal conservation data [Multiz alignment for representative vertebrate species, black lines], and distribution of predicted Fox and Fos-Jun transcription factor binding sites. Note, the heavy conservation in the Prox-3 region across all placental mammals, and in the Dist-2 region, that this is confined to the short interval detected by both Foxc1 antibodies. In addition, Prox-3 contains a cluster of 3 predicted binding sites for Foxc1, while Dist-2 includes a cluster of Foxc1 and Fos-Jun predicted binding motifs. [Placental mammal conservation by PhastCons (dark green “Placental Cons” track); placental mammal conserved elements (cherry “Placental El” track”)].

Foxc1 expression phenocopies Arhgap36-induced resistance to Smoothened inhibition and facilitates ciliary accumulation of Gli2.

(A) Quantitative western blotting demonstrates that resistance to sonidegib inhibition observed in Foxc1-expressing Gli2-mGFP NIH3T3 cells recapitulates observations in NIH3T3 cells with ectopic expression of Myc-FLAG Arhgap36 [Arhgap36-MF]. (B,C) Accumulation of Gli2-mGFP at the tips of primary cilia in Gli2-mGFP NIH3T3 cells that express Foxc1, relative to relevant empty vector control [pLXSH]. Cells serum-starved for 20 h in media containing 0% FBS, 0.5% BSA, and next to stimulated with either vehicle or Smoothened agonist (SAG). Beeswarm/box plot B shows distribution of Gli2-mGFP intensity in individual cilia across all conditions tested [boxplot: box/whiskers – quartiles; black bar – median; black dot – mean; N = 9 combined experiments]. Mean ciliary Gli2-mGFP intensity values are provided in plot C.

Foxc1-induced Gli1 expression is attributable to Arhgap36.

(A) qPCR analyses demonstrate substantial increase in Gli1 mRNA levels in NIH3T3 cells expressing Foxc1. This effect is reversed by shRNA-mediated knock-down of Arhgap36 expression.

Overall survival of neuroblastoma patients stratified by Arhgap36 expression levels and MYCN amplification status.

(A) Analysis was performed on the GSE49711, E-MTAB-1781 and TARGET 2018 data merged into a single dataset comprising n=1348 individuals (Figure 8D). MYCN-amplified cases were segregated into a separate group. Kaplan-Meier plots show: a) best overall survival for patients with high Arhgap36 expression and without MYCN amplification, b) preserved lower overall survival rate among patients with low Arhgap36 expression and without MYCN amplification (HR 1.7 – 3.4), c) predictably poor survival among high-risk MYCN-amplified cases (HR 6.1 – 11.3 compared to Arhgap36 high reference).