SLIT3-mediated intratumoral crosstalk induces neuroblastoma differentiation via a spontaneous regression-like program

Abstract

Background: Neuroblastoma, the most common pediatric extracranial solid tumor, has heterogeneous clinical outcomes ranging from malignant progression to spontaneous regression. With the highest frequency of the elusive spontaneous regression, low-risk INSS Stage 4S neuroblastoma represents an ideal model for mechanistic investigation. Spontaneous regression is often accompanied by tumor differentiation, but the mechanisms underlying this process remain largely unclear.

Methods: Single-nucleus transcriptomics (snRNA-seq) data of neuroblastoma samples were obtained from the Synapse repository to investigate the composition of heterogeneous tumor cell clusters. The feature of the Stage 4S-specific tumor cell subpopulation was revealed through differential expression analysis, pathway enrichment analysis and pseudotime analysis, followed by clinical significance validation on public cohort datasets. The biological function of secreted SLIT3 was validated using multiple in vitro models, including recombinant protein treatment, conditioned medium treatment, and cell lines coculture, to confirm the intratumoral crosstalk effect. Orthotopic and subcutaneous xenograft models were established to verify SLIT3's in vivo function. Cellular bulk RNA-seq analysis was performed with or without SLIT3 recombinant protein treatment to discover the downstream pathways activated by SLIT3, followed by validation with specific pathway inhibitors.

Results: Analysis of snRNA-seq revealed a distinct subpopulation of tumor cells within INSS Stage 4S neuroblastoma, characterized by a spontaneous regression-like program progressing toward differentiation. Activated SLIT-ROBO signaling was found in the Stage 4S-specific tumor cell subpopulation, which strongly correlated with favorable prognosis. Further investigation into the secreted ligands in SLIT-ROBO related pathways revealed that SLIT3 displayed the most potent enrichment in Stage 4S tumors and the strongest differentiation-inducing effect. In vitro experiments using recombinant SLIT3 protein, conditioned medium, and cell lines coculture consistently demonstrated the capacity of SLIT3 to induce neuroblastoma cell differentiation via intratumoral crosstalk, as evidenced by increased neurite outgrowth and elevated expression of neuronal differentiation markers. Both orthotopic xenograft and subcutaneous xenograft models demonstrated that SLIT3 expression suppressed tumor growth, leading to in vivo tumor differentiation. Mechanistically, PLCβ/PKC signaling mediates the SLIT3-induced neuroblastoma cell differentiation.

Conclusions: Stage 4S-specific tumor cell subpopulation exhibits a spontaneous regression-like program, from which SLIT3 mediates intratumoral crosstalk and promotes neuroblastoma differentiation via PLCβ/PKC signaling. These findings provide new insights into the mechanism of spontaneous regression in neuroblastoma and offer novel therapeutic targets for differentiation-based treatment strategies.

Keywords: Differentiation induction therapy; Intra-tumor cell heterogeneity; Neuroblastoma; SLIT-ROBO signaling; SLIT3.

Fig. 1

A distinct tumor cell subpopulation exists in Stage 4S neuroblastoma. A tSNE visualization displaying identified tumor cell clusters selected for subsequent analyses. B tSNE visualization of identified tumor cells colored by disease stage. C tSNE visualization of identified tumor cells colored by stage-specific clusters. D Heatmap showing GSVA gene sets enrichment scores of the top 15 pathways from GO biological process database in clusters cIV, cSHR and cIVS. E Violin plots showing neuronal function markers expression across clusters cIV, cSHR and cIVS. F Pseudotime analysis of the identified tumor cells reveals two distinct cell fate trajectories (upper panel). Distribution of cells from clusters cIV, cSHR, and cIVS along the pseudotime trajectory (lower panel). G Pseudotemporal expression progressions of neuronal function markers associated with differentiating neuroblastoma along the two cell fate trajectories. H Left panel: Heat map showing the expression of branch-dependent genes identified by BEAM. The middle of the heat map shows pre-branch cells (gray), with the left-side representing cells from Fate 1 (red), and the right-side representing cells from Fate 2 (blue). Color intensity indicates relative gene expression levels. Right panel: GO terms enriched from BEAM-identified genes from cluster 1 (corresponding to Fate 1) and cluster 2 (corresponding to Fate 2), respectively

Fig. 2

SLIT-ROBO signaling is activated in Stage 4S-specific tumor cell subpopulation. A Heatmap of DEGs between cIVS and cIV clusters. Analysis identified 4,903 DEGs (|FC|> 1.5, p < 0.05), including 2,363 upregulated and 2,540 downregulated genes. Top 200 upregulated and downregulated genes were shown. B Top 15 enriched pathways from Reactome database in cluster cIVS relative to cluster cIV. SLIT-ROBO signaling related pathways were highlighted in red. C Venn diagram depicting the intersection between upregulated genes in cIVS and The Human Secretome database, identifying 57 overlapping genes. D GO terms enrichment analysis of the 57 overlapping genes. The top 5 enriched terms in biological process, molecular function, and cellular component categories were shown. SLIT-ROBO signaling function-related axon guidance terms were highlighted in red. E Kaplan–Meier survival analysis based on the gene set score of"Axon guidance mediated by Slit Robo"in neuroblastoma cohorts Kocak (upper panel) and SEQC (lower panel). Patients were stratified into high (red) and low (blue) groups according to the gene set score. Statistical significance was determined by log-rank test. F Pearson correlation analysis among gene set scores in neuroblastoma cohorts Kocak (upper panel) and SEQC (lower panel). G, H Comparison of gene set scores in Stage 4 and Stage 4S samples from Kocak and SEQC cohorts. ***p < 0.001, Student’s t-test. I Venn diagram showing the identification of 9 candidate genes present in the gene set of Axon guidance from GO biological process database out of the 57 overlapping genes. J Volcano plot highlighting the 9 candidate genes upregulated in cluster cIVS relative to cluster cIV. Upregulated genes are colored in yellow and downregulated genes are colored in green

Fig. 3

SLIT3 promotes neuroblastoma cell differentiation through intratumoral crosstalk. A tSNE visualization of the expression distribution for the 9 candidate genes across the stage-specific clusters (upper panel), with violin plots showing expression levels in each cluster (lower panel). B Half violin plots displaying expression of the 9 candidate genes in Stage 4S versus Stage 4 tumors from neuroblastoma cohorts including Kocak (upper panel), SEQC (middle panel), and Versteeg (lower panel). Gene expressions were presented as log2 of z-score. ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001, Student’s t-test. C CellPhoneDB analysis revealing the intratumoral crosstalk within clusters cIVS and cIV, as well as their respective interactions with the cluster cSHR. D Western blot analysis of the neuronal differentiation marker MAP2 in SK-N-SH cells following overexpression of the candidate genes

Fig. 4

SLIT3 promotes neuroblastoma cell differentiation. A Representative images of SK-N-BE(2) cells treated with or without 200 ng/mL SLIT3 recombinant protein for 72 h, and statistical analysis of cell long axis length (n = 36 for Ctrl group and n = 24 for SLIT3 group, Student's t-test, ****p < 0.0001). Scale bars: 200 μm. B Representative images of SK-N-SH cells treated with or without 200 ng/mL SLIT3 recombinant protein for 72 h, and statistical analysis of cell long axis length (n = 30 for Ctrl and n = 24 for SLIT3, Student's t-test, ****p < 0.0001). Scale bars: 200 μm. C Western blot analysis of neuronal differentiation markers pTrkA, MAP2 and SNAP25 in SK-N-BE(2) cells and Kelly cells treated with or without 200 ng/mL SLIT3 recombinant protein for 72 h. D RT-qPCR analysis of neuronal differentiation marker SNAP25, CHGA, SYN3 mRNA expression in SK-N-BE(2) cells treated with or without 200 ng/mL SLIT3 recombinant protein for 72 h. Data were presented as mean ± SD (n = 3, Student's t-test, **p < 0.01, ****p < 0.0001). E RT-qPCR analysis of neuronal differentiation marker SNAP25, CHGA, SYP mRNA expression in SK-N-SH cells treated with or without 200 ng/mL SLIT3 recombinant protein for 72 h. Data were presented as mean ± SD (n = 3, Student's t-test, ***p < 0.001). F Western blot analysis of SLIT3-3F expression in wild type (wt) and stable 293T cell lines (Ctrl-3F, pLenti6-3Flag; SLIT3-3F, pLenti6-SLIT3-3Flag) for conditioned medium production. G Western blot analysis of neuronal differentiation markers pTrkA and SNAP25 in SK-N-BE(2) cells and Kelly cells treated with control (Ctrl_CM) or SLIT3-containing (SLIT3_CM) conditioned medium. H RT-qPCR analysis of neuronal differentiation marker SNAP25, CHGA, SYN3 mRNA expression in SK-N-BE(2) cells treated with control (Ctrl_CM) or SLIT3-containing (SLIT3_CM) conditioned medium for 72 h. Data were presented as mean ± SD (n = 3, Student's t-test, ***p < 0.001, ****p < 0.0001). I RT-qPCR analysis of neuronal differentiation marker SNAP25, CHGA, SYP mRNA expression in SK-N-SH cells treated with Ctrl (Ctrl_CM) or SLIT3 (SLIT3_CM) conditioned medium for 72 h. Data were presented as mean ± SD (n = 3, Student's t-test, **p < 0.01, ***p < 0.001). J Schematic diagram showing the experimental design of doxycycline-inducible cell lines coculture. K, L Western blot analysis of doxycycline-induced SLIT3-3F expression upon 1 μg/mL doxycycline treatment for 48 h in donor cell lines (TRE-3F, pLVX-TRE-3Flag; TRE-SLIT3-3F, pLVX-TRE-SLIT3-3Flag) derived from SK-N-SH cells (K) and SK-N-BE(2) cells (L). M Western blot analysis of neuronal differentiation marker pTrkA in cocultured SK-N-SH and SK-N-BE(2) recipient cells

Fig. 5

SLIT3 suppresses tumor growth and promotes neuroblastoma differentiation in vivo. A Schematic workflow illustrating the experimental design of orthotopic adrenal xenograft model. B Representative IVIS images of orthotopic adrenal xenografts in mice from Ctrl and SLIT3 groups at early (Day 12) and late stage (Day 29). C Quantification of bioluminescence signals at the presented days. Statistical significance between Ctrl group (n = 6) and SLIT3 group (n = 6) was determined by Student's t-test at each time point. Mean ± SEM. D Schematic workflow illustrating the experimental design of subcutaneous xenograft model. E Representative IVIS images of subcutaneous xenografts in mice at early (Week 5) and late stage (Week 11). F Quantification of bioluminescence signals at the presented weeks. Statistical significance between signals from Ctrl group (n = 10) and SLIT3 group (n = 10) was determined by Student's t-test at each time point. Mean ± SEM. G Representative images of H&E staining and immunohistochemical staining for SLIT3, SNAP25, and CHGA in subcutaneous xenograft tumor sections. Scale bars: 50 μm (20 ×) and 25 μm (40 ×). H Quantification of the DAB signal intensity in immunohistochemical staining of SLIT3, SNAP25, and CHGA. Statistical analyses were performed using data collected from 5 random fields of view for each protein per group. Student's t-test, **p < 0.01, ***p < 0.001

Fig. 6

PLCβ/PKC signaling mediates SLIT3-induced neuroblastoma cell differentiation. A Volcano plot displaying DEGs comparing cells treated with or without 200 ng/mL SLIT3 recombinant protein for 72 h. Upregulated genes are colored in yellow and downregulated genes are colored in green. B GSEA on the MSigDB database revealed significant enrichment of SLIT-ROBO signaling gene sets and mature neuronal synaptic function gene sets upon SLIT3 treatment C GSEA analysis shows top 10 enriched pathways from Reactome database in SLIT3-treated group compared with Ctrl group. Upregulated and downregulated pathways are colored in yellow and green, respectively. D PLCβ/DAG/IP3 signaling-related gene sets enriched in Stage 4S sample cells compared with Stage 4 sample cells within cluster cSHR (upper panel), and consistently enriched in cluster cIVS compared with cluster cIV (lower panel). E Chord diagram showing gene composition of PLCβ/DAG/IP3 signaling-related gene sets. F Heatmap showing expression of common genes from PLCβ/DAG/IP3 signaling-related gene sets in SLIT3-treated versus Ctrl groups. G A brief overview of PLCβ/DAG/IP3/PKC signaling cascade. Upon activation, membrane-bound PLCβ catalyzes PIP2 hydrolysis into DAG and IP3. DAG directly activates PKC, while free IP3 binds to IP3R receptors on endoplasmic reticulum membrane, triggering calcium release and further activating PKC. H Western blot analysis of neuronal differentiation markers pTrkA, MAP2, SNAP25 in SK-N-BE(2) cells treated with 1 μM U-73122, or 2 μM Ro 31–8220 for 72 h in the presence of Ctrl or SLIT3 conditioned medium. I RT-qPCR analysis of neuronal differentiation markers GAP43, TH, and ENO2 mRNA expression in SK-N-BE(2) cells treated with 1 μM U-73122, or 2 μM Ro 31–8220 for 72 h in the presence of Ctrl or SLIT3 conditioned medium. Data were presented as mean ± SD (n = 3, one-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Fig. 7

Schematic model of SLIT3-mediated neuroblastoma differentiation. Stage 4S-specific tumor cell subpopulation-derived SLIT3 supports intratumoral communication (left). PLCβ/PKC signaling mediates SLIT3-induced differentiation in neuroblastoma cells (right)