Super-enhancer hijacking drives ectopic expression of hedgehog pathway ligands in meningiomas

Abstract

Hedgehog signaling mediates embryologic development of the central nervous system and other tissues and is frequently hijacked by neoplasia to facilitate uncontrolled cellular proliferation. Meningiomas, the most common primary brain tumor, exhibit Hedgehog signaling activation in 6.5% of cases, triggered by recurrent mutations in pathway mediators such as SMO. In this study, we find 35.6% of meningiomas that lack previously known drivers acquired various types of somatic structural variations affecting chromosomes 2q35 and 7q36.3. These cases exhibit ectopic expression of Hedgehog ligands, IHH and SHH, respectively, resulting in Hedgehog signaling activation. Recurrent tandem duplications involving IHH permit de novo chromatin interactions between super-enhancers within DIRC3 and a locus containing IHH. Our work expands the landscape of meningioma molecular drivers and demonstrates enhancer hijacking of Hedgehog ligands as a route to activate this pathway in neoplasia.

Fig. 1. Large SCNAs mutually exclusive to 22q-loss.

Meningiomas lacking alterations in known genomic drivers, including loss of chromosome 22q, frequently exhibited somatic copy number losses in chromosomes 2q or 3p, or co-occurring whole-chromosomal gains or copy-neutral loss of heterozygosity (CN-LOH) of 16 additional chromosomes. Meningioma samples that acquired at least one of the events that are mutually exclusive to 22q-loss or 22q-loss (n = 150) are aligned horizontally while chromosomes (or chromosome arms) with SCNA types are vertically aligned. Only the events that are mutually exclusive to 22q-loss (as well as 22q-loss itself) are presented. If a chromosome name does not contain p or q arm designation such as 20-GAIN, the event covered at least 50% of both arms. Most of such gains covered >80% of the chromosome (or the arm for acrocentric chromosomes). SCNA types are colored according to the legend at the bottom right (Type). The WHO grade and histology of the samples are labeled using color bars on top. Source data are provided as a Source Data file.

Fig. 2. SCNAs affecting 2q and 7q suggest the involvement of IHH and SHH in meningioma oncogenesis.

a Distribution of SCNAs across the entire chromosome 2q arm for samples that acquired at least one somatic copy number loss on chromosome 2q are shown. Aside from a single case with POLR2A recurrent mutation (MN-64636), all tumors exhibited complex rearrangements. Event types are colored according to the legend at the top right (Type). Two dashed lines mark a 500 kilobases (kb) region from 219.65 Mb to 220.15 Mb, which will be shown in (b). The known driver alteration and WHO grade for the respective samples are shown at the right of the main panel. b The 500 kb interval, highlighted in (a), is extracted to provide a detailed view. Among the samples with complex rearrangements, we observed consistent copy number neutrality of a small genomic region that encompassed IHH and three other genes, represented by a gray-shaded area. The colors of SCNA types are the same as (a). At the bottom, Gencode basic genes (v38lift37) in this region are shown (protein coding genes in blue; used also in (c, d). For each gene, only the canonical transcript is selected. c A cluster of focal gains at 2q35 found in the remaining mutation-negative meningiomas is shown. An interval from 217.0 Mb to 220.5 Mb on chromosome 2 is depicted. In all these cases, IHH was amplified. The bottom two samples as well as a focal gain of MN-3479 were identified after applying weaker filtering conditions to copy number gain calls by ExomeCNV (GAIN-filtered, shown in brown color, see Methods). d Four of the remaining mutation-negative samples were found to harbor SCNAs on chromosome 7q arm, which encompasses the gene SHH. These SNCAs included chromothripsis, a focal gain at 7q36.3 and complex gains. The middle panel shows the copy ratio (tumor vs. normal) variation across 7q arm for MN-52454. This plot shows the whole chromosome 7q arm. Source data are provided as a Source Data file.

Fig. 3. RNA-Seq data supports Hh signaling activation in meningiomas acquiring SVs on 2q35 and 7q36.3.

a Clustering of meningiomas based on RNA-Seq data is presented using t-distributed stochastic neighbor embedding (t-SNE) of the first five eigenvectors obtained from spectral clustering. Each sample is represented by a number (for cluster membership) and color (for underlying driver event). A curve that encloses the samples belonging to the cluster 1 (named as Hh cluster) is drawn for illustration purposes. b A volcano plot of the differential expression (DE) tests for the Hh cluster against other meningiomas is shown. Red points show DE genes in the gene co-expression module M5 (n = 112). The top 20 genes ranked by s-values based on the local false sign rate (s_LFSR) are labeled. Volcano plots for other clusters are presented in Supplementary Fig. 6. c Module eigengene distribution across transcriptional clusters for the module M5 is shown. Only the Hh cluster showed significant association with M5 (linear regression using 42 biologically independent samples, Benjamini-Hochberg corrected P-value = 1.2 × 10⁻¹⁴, Supplementary Fig. 7 and Supplementary Data 14). A boxplot indicates median (middle line), the first (Q1) and the third (Q3) quartiles (box), the smallest value down to Q1–1.5 IQR and the largest value up to Q3 + 1.5 IQR (whiskers), where IQR = Q3–Q1. Values beyond the end of the whiskers are plotted individually (outliers). Points are jittered horizontally to avoid overlaps. d Size factor-normalized read counts are plotted against meningioma driver subgroups for IHH and SSH. Expression of IHH and SHH were elevated among samples with SVs on 2q and 7q, respectively. On top of each panel, s_LFSR and posterior log2 fold change (postLFC) from DE analysis (Hh vs. others) are shown. TD, tandem duplication. CTX, inter-chromosomal translocation. CNV, copy number variation. CT, chromothripsis. Source data are provided as a Source Data file.

Fig. 4. Hedgehog pathway activation in meningiomas with SVs on 2q35 and 7q36.3.

Meningiomas with predicted Hh pathway activating events, including recurrent variants in SMO and SVs on 2q35 and 7q36.3, exhibit tumor-specific staining of GAB1^,, a marker of Hh signaling activity. By contrast, we did not observe the staining of GAB1 in meningiomas with previously established non-Hh mutations. SSTR2 is a marker for meningioma cells. The light blue scale bar in each image represents 200um. An additional 5 cases are shown in Supplementary Fig. 10.

Fig. 5. Tandem duplication involving IHH alters chromatin structure and creates neo-loops between DIRC3 super-enhancers and IHH.

HiChIP data of TD samples indicates the emergence of TAD boundaries that were not found in the control set (Supplementary Fig. 13), permitting interaction of DIRC3 super-enhancers with the IHH locus. a HiChIP map, colors are based on the logarithm (base 10) of counts per million (CPM). b Significant loops shared between TD and control sets. c TAD boundaries detected by SpectralTAD. d Significant loops found only in the TD set (neo-loops). e TADs at edges of the tandem duplication segments (shown below). Their boundaries are not found in controls (Supplementary Fig. 13). It follows from the neo-loops that they form a neo-TAD such that domain B of the left copy and domain A of the right copy are juxtaposed ([A-B][A-B]) as a result of the TD. Another representation of this neo-TAD is presented in Supplementary Fig. 14. f Location of the tandem duplications for the samples included in the TD set. g Proportion of meningiomas in our H3K27ac ChIP-seq cohort (n = 37) that harbor an overlapping super-enhancer in the designated region. h Gencode basic genes (v38lift37) in this region. Genes overlapping the common super-enhancers as well as IHH are labeled. In addition, genes that exhibited ectopic expressions and were mapped upstream of IHH in tandem duplication cases are also labeled (Supplementary Fig. 9). Source data are provided as a Source Data file.