Transition to a mesenchymal state in neuroblastoma confers resistance to anti-GD2 antibody via reduced expression of ST8SIA1

We show that GD2 expression is regulated by lineage plasticity in neuroblastoma and identify ST8SIA1 downregulation as a key bottleneck to GD2 synthesis. EZH2 inhibition in mesenchymal NB cells induces an adrenergic-like state, elevates ST8SIA1 expression, and synergizes with anti-GD2 therapies.
Transition to a mesenchymal state in neuroblastoma confers resistance to anti-GD2 antibody via reduced expression of ST8SIA1

Anti-GD2 immunotherapy in the clinic

Neuroblastoma, a malignancy arising from primitive neural crest cells of the developing autonomic nervous system, is the most common extracranial solid tumor of childhood and accounts for approximately 10% of pediatric cancer deaths. Nearly all neuroblastoma patients are under the age of 5, and approximately 50% of patients with high-risk disease will relapse. The adoption of intensified, multimodal therapy has resulted in improvements in survival for children diagnosed with high-risk neuroblastoma. In particular, the utilization of the anti-GD2 antibody dinutuximab in upfront treatment protocols has reduced relapse rates, resulting in significantly improved overall survival. Dinutuximab was approved in 2015 by the FDA and is the first immunotherapy approved for pediatric cancers. Despite the success of anti-GD2 therapy, many patients still relapse and mechanisms of resistance to anti-GD2 are poorly understood.

GD2 is a disialoganglioside that is synthesized from ceramide via a complex pathway  involving the  activity of sialyltransferases and glycosyltransferases (Figure 1). GD2 integrates into the outer leaflet of the cell membrane and has been proposed to play roles in suppressing the immune system, as well as in metastatic progression. Despite being expressed on almost all neuroblastoma tumors to some degree, the mechanisms governing GD2 regulation and anti-GD2 response are not well understood. Our studies sought to better understand these pathways and identifying pharmacologic tools to augment them. 

Figure 1: Schematic showing the complete ganglioside synthesis pathway Gangliosides synthesis enzymes are italicized and protein acronyms are listed in parentheses. GD2 is synthesized from ceramide derivatives by a network of sialyltransferases and glycosyltransferases. Sialyltransferase ST8SIA1 converts GM3 to GD3, which is subsequently converted to GD2 by B4GALNT1.

GD2 expression is associated with cell state

In vitro models of neuroblastomas have been shown to exist in at least two known epigenetic states. These (nor)adrenergic and mesenchymal epigenetic states resemble developmental programs activated during neural crest development. For example, progenitor cells of the autonomic nervous system commit to adrenergic or mesenchymal tissues depending on the expression of key transcription factors such as PHOX2B or PRRX1, respectively. Neuroblastoma tumors can co-opt the mesenchymal lineage to resist cytotoxic chemotherapies, targeted drugs such as ALK inhibitors, and differentiating agents such as retinoic acid.

In our study, we identified that the mesenchymal lineage is also associated with loss of GD2 expression and decreased response to anti-GD2 therapy. We catalogued GD2 expression for 23 neuroblastoma cell lines and binned cell lines into GD2-low and GD2-high groups. Differential gene expression analysis using public RNA-sequencing available from the Cancer Cell Line Encyclopedia revealed that GD2-low cell lines were strongly correlated with a mesenchymal signature, and GD2-high cell lines were highly correlated with an adrenergic signature. Forced induction of an adrenergic-to-mesenchymal switch with ectopic expression of transcription factors PRRX1 or NOTCH3 reduced GD2 expression and response to anti-GD2 therapy. These studies demonstrate that lineage plasticity is a key regulator of GD2 expression in neuroblastoma which determines response to anti-GD2 antibody.

High GD2 surface expression is correlated with high mRNA expression of ST8SIA1

One of the most striking observations between GD2-low and GD2-high cells was the low expression of ST8SIA1 (GD3 synthase) in GD2-low cells as opposed to B4GALNT1 (GD2 synthase) (Figure 2A-B). Many studies have used B4GALNT1 expression has a surrogate of GD2 density, but our studies show that ST8SIA1 expression correlates more strongly with GD2 density. We observed downregulation of ST8SIA1 in all GD2-low models, including parental cell lines, isogenic cell lines sorted into GD2-low and GD2-high populations, and in adrenergic cells lines converted into a mesenchymal lineage. Ectopic expression of ST8SIA1/GD3 synthase, but not B4GALNT1/GD2 synthase, substantially increased surface GD2 expression and restored response to anti-GD2 therapy. 

 

Figure 2: GD2 expression correlates with expression of ST8SIA1 (GD3 Synthase). A-B) The percent of GD2-expressing cells was correlated against ST8SIA1 (A) or B4GALNT1 (B) expression for cell lines available in CCLE.

Epigenetic inhibitors increase GD2 expression

We observed that the ST8SIA1 promoter was strongly associated with increased H3K27me3 in GD2-low cell lines. H3K27me3 is catalyzed by EZH2, a core subunit of the polycomb repressive complex 2.  EZH2 inhibition promoted widespread loss of H3K27me3, induced a reprogramming of cell state into an adrenergic-like expression profile, increased GD2 expression, and the response to an anti-GD2 antibody (Figure 3A-B).

 

Figure 3: EZH2 inhibition restores GD2 expression and anti-GD2 response in a GD2-low, mesenchymal model of neuroblastoma. A) GD2 expression in SK-N-AS xenograft tumors treated with 500 mg/kg BID tazemetostat. B) Tumor growth curves showing bioluminescence of SK-N-AS tumors treated with dinutuximab, tazemetostat, or both combined.

Proposed model of GD2 regulation in neuroblastoma

We propose that transition from an adrenergic state to a mesenchymal state decreases GD2 expression through downregulation of ST8SIA1, rendering cells non-responsive to anti-GD2 immunotherapy. Treatment with an EZH2 inhibitor induces an adrenergic-like state through transcriptional, epigenetic, and chromatin remodeling, thereby re-expressing ST8SIA1, increasing GD2 expression, and restoring sensitivity to anti-GD2 antibody.

Where do we go from here?

There are a few outstanding questions from our studies.

  • There has been no evidence to date directly linking anti-GD2 therapy to the loss of GD2 expression in human tumor samples. This is in part to the lack of accessible tissue and the difficulty in assessing GD2 in relapsed tumors. Our data strongly suggest that ST8SIA1 mRNA predicts GD2 expression on the cell surface. This observation could have important implications in using ST8SIA1 as an mRNA biomarker for anti-GD2 sensitivity and response to therapy without requiring live cells to perform flow cytometry.
  • The existence of ‘mesenchymal’ cells in human neuroblastoma tumors remains controversial. Public RNA-sequencing data from neuroblastoma tumors available in Treehouse & TARGET data sets indicate that many tumors express mesenchymal-like transcriptional programs. However, more data is required to fully delineate the presence of different cell states in primary and relapsed neuroblastoma tumors, and correlating those cells to GD2 expression.
  • While all mesenchymal cell lines were GD2-low in our study, not all adrenergic cell lines were GD2-high. Low GD2-expressing adrenergic cell lines express low ST8SIA1 and respond to EZH2 inhibition, but it’s not clear the mechanism of ST8SIA1 More investigation is needed to parse out regulation of the ST8SIA1 in different lineage subtypes.
  • EZH2 inhibitors have been shown effective in increasing GD2 in Ewing sarcoma and small cell lung cancers, suggesting broad utility of PRC2 inhibitors to synergize with anti-GD2 therapy. Clinical studies will need to be performed to evaluate the combinatorial efficacy of EZH2 inhibition with anti-GD2 therapy.