Enteric neural crest differentiation in ganglioneuromas implicates Hedgehog signaling in peripheral neuroblastic tumor pathogenesis

Abstract

Peripheral neuroblastic tumors (PNTs) share a common origin in the sympathetic nervous system, but manifest variable differentiation and growth potential. Malignant neuroblastoma (NB) and benign ganglioneuroma (GN) stand at opposite ends of the clinical spectrum. We hypothesize that a common PNT progenitor is driven to variable differentiation by specific developmental signaling pathways. To elucidate developmental pathways that direct PNTs along the differentiation spectrum, we compared the expression of genes related to neural crest development in GN and NB. In GNs, we found relatively low expression of sympathetic markers including adrenergic biosynthesis enzymes, indicating divergence from sympathetic fate. In contrast, GNs expressed relatively high levels of enteric neuropeptides and key constituents of the Hedgehog (HH) signaling pathway, including Dhh, Gli1 and Gli3. Predicted HH targets were also differentially expressed in GN, consistent with transcriptional response to HH signaling. These findings indicate that HH signaling is specifically active in GN. Together with the known role of HH activity in enteric neural development, these findings further suggested a role for HH activity in directing PNTs away from the sympathetic lineage toward a benign GN phenotype resembling enteric ganglia. We tested the potential for HH signaling to advance differentiation in PNTs by transducing NB cell lines with Gli1 and determining phenotypic and transcriptional response. Gli1 inhibited proliferation of NB cells, and induced a pattern of gene expression that resembled the differential pattern of gene expression of GN, compared to NB (p<0.00001). Moreover, the transcriptional response of SY5Y cells to Gli1 transduction closely resembled the transcriptional response to the differentiation agent retinoic acid (p<0.00001). Notably, Gli1 did not induce N-MYC expression in neuroblastoma cells, but strongly induced RET, a known mediator of RA effect. The decrease in NB cell proliferation induced by Gli1, and the similarity in the patterns of gene expression induced by Gli1 and by RA, corroborated by closely matched gene sets in GN tumors, all support a model in which HH signaling suppresses PNT growth by promoting differentiation along alternative neural crest pathways.

Figure 1. Adrenergic markers CHGA (A) and DBH (B) were expressed strongly in NB and weakly in GN.

For each of these genes, immunocytochemistry produced intense labeling of NB, resembling the staining pattern of adrenal medulla (Ad). In contrast, in both GN and enteric ganglia, labeling was less intense, and was limited to a subset of neurons (red arrowheads), leaving other neurons unlabelled (blue arrowheads). Dotted white lines highlight the margins of the enteric ganglia.

Figure 2. Enteric NC markers CGRP (A), and GFAP (B) were expressed more strongly in GN than NB.

Antibodies to CGRP labeled rare cells (blue arrowheads) in NB and adrenal medulla (Ad), while labeling neurons (blue arrowheads) and processes (blue arrows) throughout both GNs and enteric ganglia (gut). Antibodies to GFAP labeled stromal cords in NB and rare adrenal medullary cells (blue arrowheads). In contrast, labeling was intense and widespread in GN and in ganglia of the gut (outlines by dotted white lines).

Figure 3. Gli1 slowed proliferation of NB cell lines while driving CGNPs to proliferate.

A) Transduction of Gli1 decreased proliferation of NB cell lines SY5Y, BE2(N), BE2(C) and BE2(S). Proliferation in SH-EP1 was relatively unaffected (P>0.05) while proliferation of CGCPs was increased 19-fold. Error bars represent standard deviation normalized mean. B) Decreased proliferation of SY-Gli cells relative to SY-GFP was confirmed by measuring the mitotic rate of both lines, expressed as PH3+ cells per 1000. Error bars represent standard deviation. C) Western blot of two independent SY-Gli and SY-GFP cultures demonstrated the N-MYC expression was not affected by Gli1 transduction, while RET was strongly induced by Gli1.