SOX11 regulates SWI/SNF complex components as member of the adrenergic neuroblastoma core regulatory

To identify the key SOX11 regulated factors and pathways contributing to the adrenergic NB cell phenotype, we performed global transcriptome analysis upon transient siRNA-mediated knockdown of SOX11 in IMR-32, CLB-GA and NGP cells after 48 h (Supplementary Fig. 4a, Supplementary Data 2).

In support of robustness and biological relevance of our experiments, we noted a significant overlap in both commonly down (n = 310, adj. pval < 0.05) and upregulated genes (n = 71, adj. pval < 0.05) across the three NB cell lines (Supplementary Fig. 4b). To further validate these findings and to filter out transcriptional bystanders, we performed an orthogonal experiment using SOX11 inducible SH-EP cells, which under control conditions do not express SOX11, and obtained transcriptome data at 9 h (SOX11 early regulated genes) and 48 h (SOX11 late regulated genes) after SOX11 induction (Fig. 4a, Supplementary Fig. 4a and c, Supplementary Data 2). First, in support of SOX11 knockdown and overexpression specificity, we observed enrichment of predicted SOX11 binding sites in the promoters of differentially expressed genes (Supplementary Fig. 4d). Second, intersection analysis yields 114 common SOX11 early regulated genes and 187 common SOX11 late regulated genes upon comparison of SOX11 knockdown versus SOX11 early (9 h) and late (48 h) regulated genes, respectively (Fig. 4b). Third, we determined which of these genes showed significant correlation with SOX11 expression in 2 independent NB tumor cohorts (GSE8504717 and GSE4554718) and narrowed down the SOX11 early and late regulated gene list and established a SOX11 early (56 genes) and late (68 genes) gene signature (Fig. 4c, Supplementary Data 3). Genes in the early SOX11 gene signature are predominantly downregulated upon SOX11 knockdown and upregulated upon SOX11 overexpression, suggesting that SOX11 acts mainly as transcriptional activator.

In order to better understand how SOX11 controls these phenotypes through its early regulated genes, we looked into top enriched gene sets in the early regulated genes which included “SWI/SNF complex” and “chromatin remodeling”. This is supported by differential early regulation of several SWI/SNF components such as BAF core component SMARCC1/BAF155 and pBAF specific PBRM1/BAF180. Further, other epigenetic regulators were found including the histone deacetylase HDAC2, the PHF6 NurD component, the H3K27me3 reader and canonical PRC1 complex component CBX2, the chromatin-modifying enzyme lysine-specific demethylase 1 KDM1A/LSD1 and pioneer transcription factor c-MYB, which were all also identified as direct SOX11 targets (see further) (Fig. 4c–d, S4e and g, Fig. 5e). Interestingly, when investigating the impact of SOX11 on all 29 known SWI/SNF components of the three known SWI/SNF complexes (c-BAF, nc-BAF, p-BAF), we observed that 20 out of 29 known SWI/SNF components are differentially upregulated after SOX11 overexpression, 16 of which are also differentially downregulated after SOX11 knockdown in at least one NB cell line. Moreover, 13 components are already differentially upregulated at the 9 h time point of SOX11 overexpression, of which 10 are directly bound by SOX11 (see further) (Fig. 4f). This strongly suggests a direct and early role of SOX11 on the expression of the SWI/SNF complex.

Taken together, the SOX11 controlled transcriptome revealed multiple functions related to cell migration, cytoskeleton, neuronal differentiation and a broad regulatory effect on multiple epigenetic regulatory factors and protein complexes.

In order to gain initial insights into the impact of the broad upregulation by SOX11 of most SWI/SNF complex components, we furthermore analysed SMARCA4 binding sites in SOX11 regulated genes and observed enrichment for SMARCA4 DNA binding only in late SOX11 regulated genes (Supplementary Fig. 5j). Although further studies are warranted, this could suggest that enhanced SWI/SNF activity associated with the observed SMARCA4 binding acts downstream of SOX11 to contribute to its induction of a gene regulatory program.

In summary, integration of SOX11 DNA binding data with SOX11-associated expression gene signatures reveals SOX11 targeting of multiple core and subunit specific SWI/SNF components as well as several other proteins implicated in diverse epigenetic regulatory processes.

SOX11 is a core regulatory circuitry transcription factor in adrenergic NB

Our SOX11 DNA binding analysis also provided further insight into proposed role of SOX11 as core regulatory circuitry (CRC) master transcription factor. CRCs are a group of interconnected auto-regulating transcription factors that form loops and can be identified by super-enhancers2,13. Our data suggest that SOX11 is a adrenergic NB CRC member. First, SOX11 binds its own promotor and binds the SOX11 3’ downstream enhancer landscape, including the above-mentioned consensus SOX11 super-enhancer (Fig. 6a). Second, binding of major CRC members including HAND2, PHOX2B and GATA3 was observed at the SOX11 promoter. Also, HAND2, GATA3, MYCN, ASCL1 and TWIST1 bind the downstream enhancer landscape including the consensus SOX11 super-enhancer (Fig. 6a). Third, SOX11 binding is observed at the promotors of HAND2, PHOX2B, GATA3, ASCL1, TWIST1, and TCF3 (Fig. 6b). Finally, as reported for other CRC transcription factor members, SOX11 knockdown causes partial CRC collapse notified by attenuated expression of several CRC members such as TCF3, ISL1, PHOX2B, TFAP2B, MYCN and KLF7. However full adrenergic-to-mesenchymal transitions and establishment of the mesenchymal CRC is not observed after SOX11 knockdown (Fig. 6c, Supplementary Fig. 6a, Supplementary Data 2), suggesting knockdown of SOX11 on its own is not sufficient to induce a lineage switch.

To investigate further the functional connection between SOX11 and the adrenergic CRC, we compared the binding sites of the major adrenergic CRC members to our own SOX11 multi-omics data. Co-binding of MYCN, ASCL1 and TWIST1 at SOX11 bound enhancers and promotors can be observed as well as HAND2, GATA3 and PHOX2B co-binding at SOX11 bound enhancers (Fig. 6d). More specifically, we observe strong co-binding of MYCN, ASCL1 and TWIST1 at transcription start sites of early and late regulated SOX11 targets (Supplementary Fig. 6b). Having established that SOX11 is a member of the adrenergic CRC, we next looked into the dynamic regulation of SOX11 and other CRC members from RNA-sequencing data obtained in a human pluripotent stem cell based differentiation model for developing human sympathoblasts (Van Haver et al., in preparation). SOX11 was found to be expressed in earlier developmental stages prior to emergence of the adrenergic master regulator PHOX2B and the other CRC members including HAND2 and GATA3 (Fig. 6e, Supplementary Fig. 6c).

Taken together, our findings support the notion that SOX11 is a canonical CRC member and plays a distinct role, during early sympathoblast development prior to emergence of the adrenergic master regulator PHOX2B and the other CRC members including HAND2 and GATA3. In conclusion, we postulate that SOX11 mediates establishment and maintenance of the adrenergic core regulatory circuitry by modulating the expression of chromatin remodeling complexes and acting as an epigenetic master regulator upstream of the core regulatory circuitry.

Discussion

Cellular mechanisms that govern lineage-specific proliferation and survival during development may be co-opted by tumor cells. Consequently, these tumors will be selectively dependent on such lineage factors offering interesting options for targeting these tumor cells whilst sparing normal tissues. We identified SOX11 as a dependency gene in adrenergic neuroblastoma (NB) which is recurrently affected by large segmental 2 p gains in high-risk NBs as well as recurrent focal gains and amplifications. SOX11 was identified as the sole protein coding gene residing in the shortest region of overlap at 2 p distal to MYCN, suggesting a role as driver for selection of the respective amplicons during tumor formation. SOX11 is known to exhibit specific expression during normal development, predominantly in the neuronal lineage. In line with the data from the CRISPR screen available through the DepMap portal, our SOX11 knockdown data support that SOX11 is a dependency factor in NB. Furthermore, higher SOX11 expression levels were found to be correlated with poor prognosis for NB patients.

To gain insight into the functional contribution to the tumor phenotype, we identified functional SOX11 target genes through genome wide DNA binding analysis combined with transcriptome profiling after SOX11 knockdown in multiple adrenergic NB cell lines and forced SOX11 overexpression in mesenchymal SH-EP NB cells. Analysis of the SOX11 regulated transcriptome identified genes implicated in epigenetic control, cytoskeleton and migration, and neurodevelopment. Subsequent identification of early transcriptionally regulated genes and direct DNA bound gene promotors revealed a remarkable enrichment for genes encoding the SWI/SNF chromatin remodeler multiprotein complex, including SMARCC1, SMARCA4 and ARID1A. In addition, several other important epigenetic regulators were noted including chromatin silencing PRC1 complex components and pioneering transcription factor c-MYB. While these targets require further individual functional validation, the finding of multiple functional targets implicated in a broad range of essential epigenetic regulatory processes is intriguing. Structural disruption of the BAF complex, achieved by silencing of its essential subunits and direct SOX11 targets ARID1A and ARID1B, exerted an epigenomic reprogramming in neuroblastoma cells, resulting in reduced neuroblastoma invasiveness and metastasis22. In line with this, overexpression of SMARCA4/BRG1, the catalytical component of the SWI/SNF complex and early direct target of SOX11, is essential for NB cell viability23 and therefore, at least partially could explain dependency of adrenergic NB cells for survival. A further independent link between SOX11 and SWI/SNF functionality comes from the observation of disrupted normal (neuronal) development in the intellectual disability Coffin-Siris syndrome24 which is mostly caused by mutations in SWI/SNF components including SMARCE1 but has also been reported in patients with SOX11 germline loss of function.

Additionally, based on integrated analysis of SOX11 occupancy as determined by CUT&RUN and transcriptome analysis upon knockdown, we propose SOX11 as a master transcription factor of the recently established adrenergic core regulatory circuitry (CRC) in neuroblastoma3,4. In addition to its role as CRC member, SOX11 may act also in an upstream hierarchical function based on gene expression analysis of SOX10 positive neural crest derived maturing sympathetic adrenergic neuroblasts. While the adrenergic master regulator PHOX2B is strongly induced at day 23 of differentiation together with several CRC TFs including HAND2 and GATA3, SOX11 is clearly expressed much earlier from day 16 on of the differentiation track. Taken together, these observations, together with the previously established role of SWI/SNF chromatin remodeling in maintenance of lineage-specific25, we hypothesize that SOX11 allows NB cells to benefit from enhanced SWI/SNF activity and chromatin remodeling to sustain the establishment and maintenance of the adrenergic core regulatory circuitry of these arrested immature transforming sympathoblasts during tumor initiation. We also propose that several SOX11 driven functions are co-opted by the transformed neuroblasts contributing to the aggressive phenotype of high-risk adrenergic NBs. Further support for this view comes from the recent description of a unique aggressive transitional cell state important for the inter-transition between adrenergic and mesenchymal cells through single-cell RNA-sequencing analysis of peripheral neuroblastic tumors26. Interestingly, SOX11 is described as a marker gene of the transitional state, amongst MYCN and others, while GATA3 and HAND2 mark the adrenergic state. This is in line with the fact that SOX11 was reported to be activated in a so-called proliferative active bridging population (transient cellular state) connecting a progenitor cell type coined Schwann cell precursors and their differentiated counterpart, chromaffin cells27, while the transitional signature in the above described tumors is enriched in the bridging population. In this context it is intriguing that depletion of direct SOX11 target ARID1A promotes partial adrenergic-to-mesenchymal CRC conversion in adrenergic NB cells by regulating enhancer mediated expression through alteration of the binding sites of the SWI/SNF chromatin remodeling complex28. Additionally, a recent paper also attributed a crucial role for SOX11 as one of the required factors for fibroblasts-to-neuron conversion in the presence of exogenous reprogramming factors29. However, while we observed an enrichment of adrenergic and mesenchymal gene signatures upon SOX11 overexpression and knockdown respectively, SOX11 overexpression or knockdown in itself was not sufficient to induce a full transition in cell lineage, at least not at 48 h after induction or knockdown of SOX11.

In conclusion, we identify SOX11 as a dependency factor and member of the CRC in adrenergic high-risk NB with a putative function as epigenetic master regulator upstream of the core regulatory circuitry. Further DNA occupancy data for SWI/SNF complex components and CRC members during normal differentiation and upon controlled perturbation in NB cells will shed more light on how this CRC is formed and how it is hijacked during tumor development. Additionally, studies in in vitro cellular models and targeted overexpression to the sympathetic adrenergic lineage in mice or zebrafish as well as developmental studies in animals or hPSC differentiation models are needed to further explore the complex interplay of the broad range of transcription factors in adrenergic neuroblasts.