Functional Genomics Identifies Tis21-Dependent Mechanisms and Putative Cancer Drug Targets Underlying Medulloblastoma Shh-Type Development

Abstract

We have recently generated a novel medulloblastoma (MB) mouse model with activation of the Shh pathway and lacking the MB suppressor Tis21 (Patched1^+/-/Tis21^KO ). Its main phenotype is a defect of migration of the cerebellar granule precursor cells (GCPs). By genomic analysis of GCPs in vivo, we identified as drug target and major responsible of this defect the down-regulation of the promigratory chemokine Cxcl3. Consequently, the GCPs remain longer in the cerebellum proliferative area, and the MB frequency is enhanced. Here, we further analyzed the genes deregulated in a Tis21-dependent manner (Patched1^+/-/Tis21 wild-type vs. Ptch1^+/-/Tis21 knockout), among which are a number of down-regulated tumor inhibitors and up-regulated tumor facilitators, focusing on pathways potentially involved in the tumorigenesis and on putative new drug targets. The data analysis using bioinformatic tools revealed: (i) a link between the Shh signaling and the Tis21-dependent impairment of the GCPs migration, through a Shh-dependent deregulation of the clathrin-mediated chemotaxis operating in the primary cilium through the Cxcl3-Cxcr2 axis; (ii) a possible lineage shift of Shh-type GCPs toward retinal precursor phenotype, i.e., the neural cell type involved in group 3 MB; (iii) the identification of a subset of putative drug targets for MB, involved, among the others, in the regulation of Hippo signaling and centrosome assembly. Finally, our findings define also the role of Tis21 in the regulation of gene expression, through epigenetic and RNA processing mechanisms, influencing the fate of the GCPs.

Keywords: Sonic Hedgehog; cerebellar precursor cell; chemokines; drug target; medulloblastoma model; neural migration; primary cilium; retina.

Figure 1

A Venn diagram showing four genotype pairwise comparisons and the intersection of their differentially expressed gene/sequences set A–D. Set A corresponds to the pairwise comparison Ptch1^+/−/Tis21^KO vs. Ptch1^+/−/Tis21^+/+; Set B refers to Ptch1^+/−/Tis21^+/+ vs. wild type; Set C concerns Ptch1^+/−/Tis21^KO vs. Ptch1^+/+/Tis21^KO; Set D represents the double-knockout contribution in background wild type.

Figure 2

A molecular overview of our Shh-deregulated mouse model, characterized by the genetic disruption of hedgehog signaling resulting from a heterozygous mutation in the negative regulators Ptch1, whose effects are enhanced by the homozygous deletion of the Tis21 tumor suppressor gene. The figure shows the proteins mainly involved in neural developmental pathways, encoded by the differentially expressed genes of set B and set D and shown in their subcellular compartments. Each differentially expressed pathway object is labeled with a thermometer that indicates the gene expression changes: downward thermometers have a blue color and indicate down-regulated expression, whereas upward thermometers have a red color and indicate up-regulated expression. The thermometer number 1 is related to the pairwise comparison Ptch1^+/−/Tis21^+/+ vs. wild type Set B, while the thermometer number 2 is related to the pairwise comparison Ptch1^+/−/Tis21^−/− vs. wild type Set D. Signaling cross-talks between developmental pathways involved in pre-neoplastic GCPs development are initiated by different growth factors and cytokines, the most part of which interacts with G-protein coupled receptors. This is supported by a strong up-regulation of many members of heterotrimeric G-proteins, their target molecules and regulators (data not shown in figure). Among those reported in literature as involved in MB (Roussel and Hatten, 2011), developmental signaling pathways appear to be up-reregulated in our experimental data, according to the far fewer percentages of negative signature genes (Chen et al., 2013) and the over-representation of WNT and axonal guidance genes present in human MB Shh-type (Northcott et al., 2011). Interestingly, a β-Catenin-Gli1 balanced interaction has been recently reported to regulate Shh-driven MB tumor growth in Ptch1 heterozygous mice in vitro (Zinke et al., 2015), while Sox7, a transcription factor that is known to reduce Wnt/β-Catenin stimulated transcription in a dose-dependent manner (Takash et al., 2001), is up-regulated in Set B and Tis21 ablation enhances its up-regulation in Set D (data not shown in figure). Furthermore, Smo-dependent non-canonical Shh pathways (Jenkins, 2009), which have been reported to modulate cytoskeleton-dependent processes and fluctuation of Ca²⁺ through the plasma membrane in mammalian neurons (Brennan et al., 2012) and suggested in possible association with MB (Briscoe and Therond, 2013), are put in light here for the first time as related to the MB Shh-type mouse model. Evidences of a deregulated Slit-Robo pathway, which is implicated in neuronal migration (Wong et al., ; Marillat et al., 2004), are present in our data with the up-regulation of the axon guidance receptor Robo4. The ligand of Robo4, Slit2, has been linked to the inhibition of MB cell invasion (Werbowetski-Ogilvie et al., 2006). Proteins belonging to the ubiquitin-dependent degradation of GCPs cell cycle regulators [24] have their genes up-regulated in our model, in particular a number of ubiquitin-conjugating enzymes and some constituents of the SCF (Skip1, Cullin1, F-box)-E3 ubiquitin ligase complex. Among them, a substrate recognition component of the SCF-type E3 ubiquitin ligase, the F-box protein Fbw7, which has been linked to a premature migration of GCPs in conditional Fbw7-knockout mice [30]. An up-regulation genes coding for proteins involved in palmitoylation (i.e., HHAT) and transport of Shh (i.e., DISP1) is noticed in in Set D, where Ptch1 sterol-sensing domain seems to control Smoothened activity through Ptch1 vesicular trafficking [34]. Retinoblastoma-associated protein (Rb), as well as its downstream effectors E2F3 and E2F5, has its correspondent gene up-regulated in set D, where the deregulation of the Rb/E2F tumor suppressor complex in MB Shh-driven has been already associated to the E2F1-dependent regulation of lipogenic enzymes in primary cerebellar granule neuron precursors (Bhatia et al., 2011). Figure 4 below shows the set of symbols whereby network objects and interactions between objects are indicated in this figure.

Figure 3

Drug targets belonging to the Set A discussed in the main text. Each gene product is labeled with a thermometer indicating the gene expression changes: downward thermometers have a blue color showing down-regulated expression, whereas upward thermometers have a red color showing up-regulated expression. The most part of the figure objects are deregulated also in other two pair comparisons. For this reason, the thermometer number 1 is related to the pairwise comparison Ptch1^+/−/Tis21^+/+ vs. wild type or Set B, the thermometer number 2 is related to the pairwise comparison Ptch1^+/−/Tis21^−/− vs. wild type or Set D, while the thermometer number 3 is related to the pairwise comparison Ptch1^+/−/Tis21^KO vs. Ptch1^+/−/Tis21^+/+ or Set A. See Figure 4 for the set of symbols, objects and interactions between objects indicated in this figure.

Figure 4

The figure shows symbols, objects and interactions between objects indicated in Figures 2, 3.

Figure 5

Drug targets percentage per functional class of the Set A deregulated genes. The drug targets identified in Set A are showed as percentage with respect to the functional classes to which they belong.