The Wnt signaling receptor Fzd9 is essential for Myc-driven tumorigenesis in pancreatic islets

Zacarías-Fluck, et al identify and validate the Wnt receptor Fzd9 as a key effector of Myc-Wnt signaling cross-talk in a mouse model of Myc-driven pancreatic insulinomas.


Introduction
Myc is a highly pleiotropic transcription factor that governs cell expansion by coordinating diverse cellular processes, including proliferation, dedifferentiation, biosynthetic metabolism, cell growth, and angiogenesis (Dang, 2012). Using the Myc-switchable pIns-MycER TAM ;RIP7-Bcl-x L β-cell tumor model in an unbiased reversible kinetic expression analysis, we previously identified genes required for maintenance of Myc-driven β-cell tumors. We found a relatively small subset of Myc target genes whose change in expression (whether induction or repression) was dependent upon sustained Myc activity, and that such changes were reversed by deactivation of Myc and consequent tumor regression. One of these candidate tumor maintenance genes was Fzd9, a member of the "frizzled" gene family of Wnt receptors (Lawlor et al, 2006). Fzd9 was significantly induced within 4 h of acute Myc activation in vivo, its expression was maintained as long as Myc activity was sustained, and rapidly decreased following Myc deactivation and tumor regression (Lawlor et al, 2006). In addition, Fzd9 has been identified as a direct Myc target gene through expression and ChIP analysis (Lawlor et al, 2006;Sabò et al, 2014).
Fzd9 is up-regulated in several types of human cancers including human gastric cancer (Kirikoshi et al, 2001), osteosarcoma  and astrocytoma (Zhang et al, 2006). Knockdown of Fzd9 has also been shown to inhibit cell proliferation and motility in hepatocellular carcinoma cell lines (Fujimoto et al, 2009). However, as Fzd9 has also shown some tumor suppressor activity in acute myeloid leukemia (Zhang et al, 2016) and non-small cell lung cancer, its role in tumorigenesis is still controversial (Winn et al, 2005(Winn et al, , 2006. In this study, we explored the role of Fzd9 in mediating and maintaining Myc oncogenic function in pIns-MycER TAM ;RIP7-Bcl-x L β−cell tumors. We show that Fzd9 does, indeed, play a critical function in the development of Myc-driven insulinomas. We, thus, identify a novel link connecting Myc and the Wnt pathway, which appears to be a required effector of Myc oncogenic activity in β-cell tumorigenesis. This implies the existence of a positive feedback loop, where Wnt/β-catenin signaling activates Myc (He et al, 1998), and Myc in turn activates Wnt/β-catenin signaling through Fzd9.
To exclude the possibility that Fzd9 deficiency inhibits MycER expression or stimulates its degradation, we monitored the presence of MycER immunohistochemically in β-cells, with an anti-ER antibody. As expected, MycER expression was undetectable in control MycER transgene-negative islets of Fzd9 KO/KO ;BclXL mice, but was confirmed to be consistently expressed in tamoxifen-treated Fzd9 WT/WT ;MycER; BclXL islets and was even higher in Fzd9 KO/KO ;MycER;BclXL islets ( Fig 1D). Hence, MycER is readily detectable in Fzd9-deficient cells, and therefore lack of Fzd9 impairs Myc-induced pancreatic β-cell expansion by mechanisms unrelated to suppression of MycER expression.
Next, to ascertain whether Myc retains its capacity to induce apoptosis in pancreatic β-cells in the absence of Fzd9, MycER was continuously activated for 3 d in the β-cells of Fzd9 WT/WT ;MycER and Fzd9 KO/KO ;MycER mice, without co-expression of BclXL. As seen for proliferation, shrinkage of islets resulting from β-cell apoptosis (previously shown in Pelengaris et al [2002]) was observed in both Fzd9-proficient and Fzd9-deficient islets ( Fig 2C). Hence, Fzd9 is not required for Myc-induced β-cell apoptosis.
Finally, additional Myc-dependent phenotypes previously described (Soucek et al, 2007), such as recruitment of mast cells ( Fig  S2A) and induction of angiogenesis ( Fig S2B) were also observed after 3 d of tamoxifen treatment in Fzd9 KO/KO ;MycER;BclXL.
Together, these observations indicate that, in the absence of Fzd9, MycER is still functionally active upon tamoxifen treatment.
Immunohistochemical staining for the proliferation marker Ki67 demonstrated that the proliferative arrest observed in Fzd9 KO/KO ;MycER; BclXL islets is in fact a progressive phenomenon already evident at 1 wk of sustained Myc activation and essentially complete by 3 wk (Fig 2F).
These results suggest that even though proliferation is initially triggered upon Myc activation in Fzd9 KO/KO ;MycER;BclXL cells, Myc's tumorigenic potential is lost over time due to the lack of Fzd9.
Fzd9 is necessary for MycER-dependent dedifferentiation and transformation of β-cells In MycER;BclXL mice, partial Myc-induced dedifferentiation of β-cells has been previously noted as one of the pleiotropic effects of oncogenic Myc, responsible for tumorigenesis and expansion of the islets of Langerhans (Pelengaris et al, 2002). To verify if such an effect was Fzd9dependent, we looked at the expression of both insulin and glucagon (markers of β-cells and α-cells, respectively). As expected, we detected high insulin expression in control MycER transgene-negative islets (Fzd9 KO/KO ;BclXL) ( Fig 3A). In contrast, insulin levels were somewhat repressed upon 3-wk tamoxifen treatment of Fzd9 WT/WT ;MycER;BclXL mice because of Myc-induced partial dedifferentiation of β-cells ( Fig  3B) (Pelengaris et al, 2002). Intriguingly, however, high levels of insulin expression were maintained in islets from Fzd9 KO/KO ;MycER;BclXL mice after tamoxifen treatment ( Fig 3C and D), indicating blunting of Mycinduced β-cell dedifferentiation in the absence of Fzd9. Indeed, pockets of fully differentiated cells can also be appreciated by hematoxylin and eosin (H&E) staining, already after 2 wk of MycER activation in Fzd9 KO/KO ; MycER;BclXL but not in Fzd9 WT/WT ;MycER;BclXL ( Fig S3).
Glucagon positive α-cells displayed their typical proportion (~10% of the total islet cellular content) and distribution at the islet periphery in Fzd9-deficient mice without the MycER transgene (Fzd9 KO/KO ;BclXL) ( Fig  3A). In contrast, in tamoxifen-treated Fzd9 WT/WT ;MycER;BclXL mice, Myc caused the appearance of α-cells scattered throughout the whole islet ( Fig 3B). Notably, Myc did not trigger this relocalization in the absence of Fzd9 (Fzd9 KO/KO ;MycER;BclXL mice treated with tamoxifen), where the α-cells displayed a more physiological distribution, although in a higher proportion than in control islets (Fig 3C and E).
Overall, these results suggest that Fzd9 is required for Myc to induce profound and long-lasting dedifferentiation in β-cells, as part of its tumorigenic program.
Gene expression analysis shows down-regulation of genes involved in β-cell survival and stress response in the absence of Fzd9 To identify potential transcriptional changes that could explain the anti-tumorigenic effect observed in the absence of Fzd9, Fzd9 WT/WT ; MycER;BclXL and Fzd9 KO/KO ;MycER;BclXL mice were treated with tamoxifen for 3 d and the islet RNA was used for microarray  analysis. This early 3-d time-point was selected to reveal those differences in gene expression that could be the cause, and not the consequence, of the subsequent phenotypic changes (such as decrease proliferation rate or dedifferentiation of β-cells). Statistical analysis of the microarray (P < 0.05 and fold-change ± 1.2) identified 933 differentially expressed genes out of 27,747 probes ( Fig 4A and Table S1). Among the genes whose expression is most decreased in Fzd9 KO/KO pancreatic islets compared with Fzd9 WT/WT expression (fold-change < −2.5), we found several early response genes, comprising transcription factors and other cellular mediators  of β-cell survival and stress response. JUNB and ATF3, for instance, are known to coordinate a β-cell survival pathway during inflammation (Gurzov et al, 2012), whereas NPAS4 is an important early mediator of β-cell stress response (Sabatini et al, 2013). Moreover, RGS2 (Dong et al, 2017) and IRS2 (Blandino-Rosano et al, 2016) regulate β-cell survival and mass.
Notably, GSEA also showed that several Myc-related gene sets are significantly down-regulated ( Fig 4C and Table S3) along with Myc-related transcriptional programs like cell cycle, metabolism, and apoptosis ( Fig  S4A-C). The fact that the absence of only one Myc target gene is able to affect the expression of a number of Myc-related signatures reveals Fzd9 as a key effector of Myc-driven reprogramming of pancreatic β-cells.
We validated the transcriptional regulation in pancreatic islets observed in the GSEA by performing qRT-PCR of some of the identified differentially regulated genes after 3 d of tamoxifeninduced MycER activation in Fzd9 KO/KO and Fzd9 WT/WT mice ( Fig 4D). We observed that, consistent with the microarray data, GPC1 (belonging to Developmental Biology Gene Set), RPL34, SRSF7 and POLD2 (MYC Targets Gene Set), were significantly down-regulated. This confirms-using a distinct quantitative method-the Fzd9mediated transcriptional regulation of several MycER targets previously observed by microarray analysis.
We complemented this validation approach by performing immunofluorescence stainings for other down-regulated MYC target genes (namely, MCM5 and proliferating cell nuclear antigen [PCNA]) on pancreatic islets after 1 wk of MycER activation and confirmed their reduced expression in the islets of Fzd9 knockout mice (Fig 4E).

Wnt signaling is engaged by Fzd9 and acts as a downstream effector of Myc-induced tumorigenesis
Given the role of Fzd9 in Wnt signaling, this pathway seemed a priori a good candidate that could contribute to the multiple Myc-dependent phenotypes observed when MycER is activated. In fact, Wnt signaling has been shown to regulate pancreatic β-cell proliferation (Rulifson et al, 2007). However, as previously noted, there was no detectable difference in the histology of pancreata from Fzd9 KO/KO and Fzd9 WT/WT mice (Fig S1), indicating that the absence of this receptor is not rate limiting at least during normal development of the organ. Nevertheless, in MycERactivated islets, there are significant differences in Wnt-related gene sets between Fzd9 knockout versus wild-type cells (Fzd9 KO/KO ;MycER;BclXL versus Fzd9 WT/WT ;MycER;BclXL) ( Fig 4F). Moreover, qRT-PCR analysis of AXIN1 and JUNB, genes belonging to WNT and β-Catenin Gene Sets, respectively, confirmed their down-regulation, although they did not reach statistical significance (P = 0.11 and 0.099, respectively; Fig S4E). In addition, immunofluorescence staining for β-Actin, part of Labbe WNT3A Targets up Gene Set (Fig 4F), showed that in the absence of Fzd9 expression, β-Actin levels are significantly lower in pancreatic islets (Fig S4D).

Discussion
A vast amount of data in the literature points at Wnt signaling as one of the major culprits in solid and liquid tumors (Zhan et al, 2017) and Wnt quantification in the islets. Total β-catenin intensity per islet area is shown. (H) Percentage of Ki67-positive cells in individual islets from Fzd9 WT/WT ;MycER;BclXL and Fzd9 KO/KO ;MycER;BclXL after 1 wk of tamoxifen treatment. An additional group of the Fzd9 WT/WT ;MycER;BclXL mice was pretreated for 2 d with the Wnt inhibitor C59 and then received both C59 + tamoxifen during 1 wk. Data information: in (D, G, H), data are represented as mean ± SD, whereas in (E), as median and interquartile range. (D, E, G) Statistical significance of differences was examined using t test (D), Mann-Whitney U test (E) and Tukey's test (G). *, ** and *** indicate P-values below 0.05, 0.01 and 0.0001, respectively. (E, G) Scale bars: 50 μm in (E), 100 μm in (G). FDR, false discovery rate; NES, normalized enrichment score. pathway inhibition via the targeting of Frizzled receptors has been suggested as a potential strategy to decrease growth and tumorigenicity of human tumors (Gurney et al, 2012). Our data indicate that simply inhibiting one receptor at a time (namely Fzd9) could be sufficient to achieve a significant therapeutic impact at least in some tumorigenic contexts, with the advantage of reducing potential side effects associated with the simultaneous inhibition of multiple Frizzled family members. Whereas Myc expression has been traditionally placed downstream of the Wnt signaling pathway, as previously discussed, others have indicated otherwise (Cowling & Cole, 2007). Here, we demonstrate for the first time the key role of the Wnt receptor Fzd9 and Wnt signaling in Mycinduced insulinomas. In this context, whereas previous results suggest that targeting Fzd9 may not be the best strategy for cancer therapy because of its dual pro-and anti-tumorigenic character (Zeng et al, 2018), our data indicate that it might be at least a valuable target when up-regulated in Myc-driven malignancies. In addition, our results suggest the existence of a novel positive feedback loop between Myc and Wnt signaling: deregulation of Myc might enhance Wnt signaling by upregulation of Fzd9, which would, at the same time, promote Myc overexpression/deregulation-because Myc is a bona fide downstream target of Wnt-thus feeding again Wnt signaling through Fzd9. It should be noted that both our genetic and pharmacological approach to delete Fzd9 and interfere with Wnt signaling, respectively, are systemic and not specific for the islets of Langerhans only. Hence, we cannot exclude that such a positive feedback loop between Myc and Wnt can also have cell autonomous implications.
The identification of Fzd9 as a key Myc-driven tumorigenic effector is particularly relevant in view of the fact that, even if Myc is one of the "most wanted" cancer targets, several direct, and indirect Myc inhibition strategies have failed because of lack of efficacy and high toxicity derived from low specificity (Whitfield et al, 2017). Thus, identification of key tumorigenic effectors and their selective targeting represents an alternative strategy to achieve high therapeutic impact in Myc-deregulated malignancies. Here, we have described a good example of this alternative approach.

Materials and Methods
Generation and maintenance of genetically engineered mice pIns-MycER TAM ;RIP7-Bcl-x L and Fzd9 KO/KO mice have been previously described (Pelengaris et al, 2002;Zhao et al, 2005). All the animal studies were performed in accordance with the ARRIVE guidelines and the 3 Rs rule of Replacement, Reduction and Refinement principles. Animals were maintained and treated in accordance with protocols approved by the CEEA (Ethical Committee for the Use of Experimental Animals) at the Vall d'Hebron Institute of Oncology. Mice (both males and females) between 8 and 12 wk old were used.

Preparation and administration of tamoxifen and C59
Tamoxifen (Sigma-Aldrich) was dissolved in peanut oil (Thermo Fisher Scientific) at 10 mg/ml. Aliquots of 1 ml were prepared and frozen at −20°C. This injectable solution was administered to mice by intraperitoneal injection (6 μl/g) every 24 h. 1 ml-syringes and 27 G needles were used for injection.
The Wnt inhibitor C59 was formulated in PEG400 (Sigma-Aldrich) at 2 g/l. C59 in PEG400 was aliquoted and stored at 4°C. Right before each treatment, an equal amount of water (1:1) was added to make a final concentration of 1 g/l of C59 in 50% PEG400. Mice were then treated with a daily dose of 10 mg/kg by oral gavage for seven consecutive days. A mixture 1:1 dilution of water and PEG400 was used as vehicle for C59-untreated animals.

Immunostaining of pancreas sections
18 h after the last administration of tamoxifen, mice were euthanized with CO 2 and pancreata collected. For BrdU staining, 150 μl of BrdU (Sigma-Aldrich) at 5 mg/ml were intraperitoneally injected 2 h before euthanasia.

Microarray analysis of pancreatic islets
Genome-wide expression analysis was performed in isolated pancreatic islets. Briefly, tamoxifen-treated Fzd9 WT/WT ;MycER;BclXL and Fzd9 KO/KO ;MycER;BclXL (n = 4) mice were euthanized by cervical dislocation. Pancreata were inflated with Collagenase P (6 ml/mouse at 0.7 mg/ml) (Roche) in HBSS injected through the bile duct. Tissues were transferred to vials containing 5 ml of Collagenase P and incubated for 20 min at 37°C with gentle shaking. Digested pancreata were poured into 50 ml tubes and washed with cold HBSS by filling the tube, performing a short spin up to 652g and removing the supernatant. Pellets were resuspended in 5 ml and exocrine tissue further removed by filtering the suspension through 100 μm restrainers. Then, tissue remaining in the filter was placed in a 6 cm plate in 4 ml of cold HBSS. Pancreatic islets were visualized with the help of a magnifier by addition of dithizone at 0.02 g/l, hand-picked and transferred into a clean 1.5 ml tube.
RNA from islets was isolated, DNAse-treated and quality assessed through Agilent 2100 Bioanalyzer. RNA was reversetranscribed to generate cDNA. Microarray was performed using a Mouse Gene Array 2.1 ST (Affimetryx). GSEA was performed using publicly available software provided by the Broad Institute (version 3.0) with the Hallmarks, Curated, Motif, gene ontology, Oncogenic Signatures and Immunological Signatures gene sets from the MsigDB (Subramanian et al, 2005). The number of permutations was set to 1,000 and the genes were ranked according to Signal2Noise. Heat map was generated using Heat mapper (Babicki et al, 2016). Hierarchical clustering was performed applying complete linkage method and based on Euclidean distance.
qRT-PCR validation of microarray qRT-PCR validation was performed in isolated pancreatic islets. Briefly, Fzd9 WT/WT ;MycER;BclXL (n = 3) and Fzd9 KO/KO ;MycER;BclXL (n = 3) were treated with tamoxifen for 3 d and pancreatic islets were then isolated from them following the protocol described in the previous section. RNA was then extracted from the samples using RNeasy kit (QIAGEN) and quantified using NanoDrop. Equal amounts of RNA were then DNAse-treated (NEB) and reverse transcribed to generate cDNA using iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad). SYBR green qRT-qPCR analysis was then performed on these cDNA samples with PerfeCTa SYBR Green FastMix, Low Rox (Quantabio) using QuantStudio 6 FLEX system (Applied Biosystems). The data thus obtained were analyzed following the comparative (ΔΔCT) method described in Livak and Schmittgen (2001). B2M (Beta-2-microglobulin) was used as the housekeeping gene. Based on our hypothesis and the data from the microarray, statistical analysis was performed using one-tailed t test. Sequences of primers used are listed in Table S4.

Image and statistical analysis
Islet size was quantified using ImageJ. Immunofluorescence (BrdU, Insulin, Glucagon, β-catenin, MCM5, PCNA, β-Actin) and Immunohistochemistry (Ki67) stainings were quantified using ImageJ and Qupath (Bankhead et al, 2017), respectively. All data were represented and analyzed using GraphPad Prism 6. One-way ANOVA and Tukey's test, or Kruskal-Wallis followed by Dunn's test were used to assess statistical significance when analyzing three groups of parametric or nonparametric distributions. Analysis of two groups were performed with t test or Mann-Whitney. For each experiment, at least three animals per group were used and multiple islets quantified. Results are shown as mean ± SD or median and interquartile range, accordingly. Statistical significance was considered when P < 0.05.

Data Availability
Microarray data were deposited under accession number GSE167073.