Long-Pentraxin 3 Affects Primary Cilium in Zebrafish Embryo and Cancer Cells via the FGF System

Primary cilium drives the left-right asymmetry process during embryonic development. Moreover, its dysregulation contributes to cancer progression by affecting various signaling pathways. The fibroblast growth factor (FGF)/FGF receptor (FGFR) system modulates primary cilium length and plays a pivotal role in embryogenesis and tumor growth. Here, we investigated the impact of the natural FGF trap long-pentraxin 3 (PTX3) on the determination of primary cilium extension in zebrafish embryo and cancer cells. The results demonstrate that down modulation of the PTX3 orthologue ptx3b causes the shortening of primary cilium in zebrafish embryo in a FGF-dependent manner, leading to defects in the left-right asymmetry determination. Conversely, PTX3 upregulation causes the elongation of primary cilium in FGF-dependent cancer cells. Previous observations have identified the PTX3-derived small molecule NSC12 as an orally available FGF trap with anticancer effects on FGF-dependent tumors. In keeping with the non-redundant role of the FGF/FGR system in primary cilium length determination, NSC12 induces the elongation of primary cilium in FGF-dependent tumor cells, thus acting as a ciliogenic anticancer molecule in vitro and in vivo. Together, these findings demonstrate the ability of the natural FGF trap PTX3 to exert a modulatory effect on primary cilium in embryonic development and cancer. Moreover, they set the basis for the design of novel ciliogenic drugs with potential implications for the therapy of FGF-dependent tumors.


Introduction
Primary cilia are antenna-like organelles protruding from most mammalian cells [1] to act as chemosensors and mechanosensors for external stimuli [2]. The primary cilium is composed by an axoneme made by microtubule triplets anchored at the cell by a basal body [3]. The basal body derives from the mother centriole of the centrosome [4], which is fundamental to cell division. The cilium is resorbed prior to mitosis to release the centrioles. Then, ciliogenesis occurs shortly after cytokinesis has been completed [3].
Recently, the zebrafish embryo has been used as a model for the study of primary cilia functions in development and diseases [5,6]. During embryonic development, primary cilia in the embryonic node are involved in the left-right asymmetry process [7] and genetic defects in primary cilia are associated with a variety of pathological conditions that are grouped under the name "ciliopathies" (reviewed in [8][9][10]). Due to their role in the modulation of various signaling pathways, including Hedgehog (Hh) and Wnt, dysregulation of primary cilia plays an important role also in cancer progression [11][12][13].

Temporal and Spatial Expression of ptx3b during Zebrafish Development
The expression of ptx3b was analyzed at different stages of zebrafish embryo development by RT-PCR and whole-mount in situ hybridization (WISH). As shown in Figure 1A, ptx3b expression, detectable in the ovary, is absent at the four-cell stage, increases during epiboly, and remains constant from the five-somite stage (ss) to the 72 h post-fertilization (hpf) stage. During somitogenesis, the expression of ptx3b is restricted to the pronephric duct primordia where it was observed up to 48 hpf ( Figure 1D-J). In addition, ptx3b is expressed in a transient manner at 26 hpf also in the notochord ( Figure 1F,G), as highlighted by the analysis of paraffin-embedded transverse cross sections of the embryo trunk ( Figure 1H), to be lost at 48 hpf ( Figure 1I,J).

Temporal and Spatial Expression of ptx3b during Zebrafish Development
The expression of ptx3b was analyzed at different stages of zebrafish embryo development by RT-PCR and whole-mount in situ hybridization (WISH). As shown in Figure 1A, ptx3b expression, detectable in the ovary, is absent at the four-cell stage, increases during epiboly, and remains constant from the five-somite stage (ss) to the 72 h post-fertilization (hpf) stage. During somitogenesis, the expression of ptx3b is restricted to the pronephric duct primordia where it was observed up to 48 hpf ( Figure 1D-J). In addition, ptx3b is expressed in a transient manner at 26 hpf also in the notochord ( Figure 1F,G), as highlighted by the analysis of paraffin-embedded transverse cross sections of the embryo trunk ( Figure 1H), to be lost at 48 hpf ( Figure 1I,J).

ptx3b Knockdown Causes Defects in the Determination of Left-Right Asymmetry in Zebrafish
In order to assess the role of ptx3b on zebrafish embryo development, we used an antisense morpholino (MO) knockdown approach. To this purpose, a splicing MO (ptx3b MO) was designed to target the exon 2/intron 2 border of the ptx3b transcript. A five-mismatch nucleotide MO, unable to bind the ptx3b mRNA (ctrl MO), was used as control. As shown in Figure 2A, RT-PCR analysis performed at 28 hpf confirmed the targeting efficacy of the ptx3b MO that caused the skipping of exon 2 in the ptx3b transcript of embryo morphants when compared to controls. Based on these results, the dose of 0.6 pmol/embryo was considered as the optimal dose of ptx3b MO to be used for further studies.
When compared to controls, 40% of ptx3b morphants showed no morphologic alterations and 50% of them exhibited only moderate defects ( Figure 2B,C). Given the hypothesis that the FGF trapping activity of PTX3 may result in primary cilium alterations and consequent defects in the leftright asymmetry process [33], WISH analysis was performed to investigate the expression of laterality genes and the positioning of visceral organs in control and ptx3b morphants. As shown in Figure  2D,E, the majority of 16-ss and 22 hpf ptx3b morphants showed bilateral, right-sided, or absent expression of spaw and pitx2 transcripts, two Nodal-related genes normally expressed at the left side of the embryo [34].

ptx3b Knockdown Causes Defects in the Determination of Left-Right Asymmetry in Zebrafish
In order to assess the role of ptx3b on zebrafish embryo development, we used an antisense morpholino (MO) knockdown approach. To this purpose, a splicing MO (ptx3b MO) was designed to target the exon 2/intron 2 border of the ptx3b transcript. A five-mismatch nucleotide MO, unable to bind the ptx3b mRNA (ctrl MO), was used as control. As shown in Figure 2A, RT-PCR analysis performed at 28 hpf confirmed the targeting efficacy of the ptx3b MO that caused the skipping of exon 2 in the ptx3b transcript of embryo morphants when compared to controls. Based on these results, the dose of 0.6 pmol/embryo was considered as the optimal dose of ptx3b MO to be used for further studies.
When compared to controls, 40% of ptx3b morphants showed no morphologic alterations and 50% of them exhibited only moderate defects ( Figure 2B,C). Given the hypothesis that the FGF trapping activity of PTX3 may result in primary cilium alterations and consequent defects in the left-right asymmetry process [33], WISH analysis was performed to investigate the expression of laterality genes and the positioning of visceral organs in control and ptx3b morphants. As shown in Figure 2D,E, the majority of 16-ss and 22 hpf ptx3b morphants showed bilateral, right-sided, or absent expression of spaw and pitx2 transcripts, two Nodal-related genes normally expressed at the left side of the embryo [34].
Next, we examined the positioning of heart, liver, and pancreas in control and ptx3b MO-injected embryos by WISH analysis using the tissue-specific probes cmlc2, prox1, and islet1, respectively. As shown in Figure 2F,G, the cmlc2 + cardiac jogging and looping processes, which occur at 26 and 48 hpf, respectively, were absent in the majority of ptx3b morphants. Moreover, ptx3b down-modulation caused significant alterations in the positioning of islet1 + dorsal pancreatic bud and of prox1 + liver when assessed at 24 hpf and 48 hpf, respectively ( Figure S3). bp band in ptx3b MO-injected embryos, which confirms the occurrence of exon skipping. ef1a serves as control; (B) percentage of embryos showing normal, moderate, or severe phenotype at 72 hpf after the injection of 0.6 pmoles of ctrl MO or ptx3b MO, respectively; (C) representative bright field whole mount pictures of the phenotypes observed in ctrl and ptx3b morphants; (D) WISH representative pictures of the alterations of the expression of the laterality genes spaw and pitx2 observed in ptx3b morphants at 16 ss and 22 hpf, respectively; (E) percentage of embryos with normal or altered expression of spaw and pitx2 in WT (untreated) and ptx3b MO-injected embryos. Data are the mean of 4 and 3 independent experiment, respectively; (F) WISH representative pictures of cmlc2 expression during normal and altered cardiac jogging and looping at 26 and 48 hpf, respectively; (G) percentage of embryos with normal or altered cardiac jogging and looping in ctrl MO and ptx3b MO-injected embryos. Data are the mean of 2 and 3 independent experiments, respectively. * p < 0.05, *** p < 0.001, ANOVA. n, total number of analyzed embryos. Next, we examined the positioning of heart, liver, and pancreas in control and ptx3b MO-injected embryos by WISH analysis using the tissue-specific probes cmlc2, prox1, and islet1, respectively. As shown in Figure 2F,G, the cmlc2 + cardiac jogging and looping processes, which occur at 26 and 48 hpf, respectively, were absent in the majority of ptx3b morphants. Moreover, ptx3b down-modulation caused significant alterations in the positioning of islet1 + dorsal pancreatic bud and of prox1 + liver when assessed at 24 hpf and 48 hpf, respectively ( Figure S3).

ptx3b Modulates Primary Cilium Length Determination via the FGF/FGFR System in Zebrafish
Primary cilia of the embryonic node, named Kupffer's vesicle (KV) in zebrafish [5], are involved in the determination of the left-right asymmetry during development [7,35]. On this basis, the length of KV primary cilia was measured in control and ptx3b morphants after acetylated α-tubulin immunostaining [6]. As shown in Figure 3, primary cilia of the KV are significantly shorter in ptx3b morphants when compared to control animals. At variance, no difference in the number of KV primary cilia was observed in ptx3b MO-injected embryos versus controls (mean ± S.E.M. equal to 21.5 ± 2.7 (n = 30) versus 22.7 ± 2.0 (n = 35) cilia per KV, respectively). To confirm the specificity of the ptx3b MO effects, one-cell stage embryos were co-injected with the ptx3b MO and an excess of hPTX3 mRNA. As shown in Figure 3D,E, hPTX3 mRNA was able to rescue the shortening of KV primary cilia caused by the ptx3b MO. Accordingly, rescued embryos showed a reduced percentage of cardiac looping defects when compared to ptx3b morphants ( Figure 3F). Primary cilia of the embryonic node, named Kupffer's vesicle (KV) in zebrafish [5], are involved in the determination of the left-right asymmetry during development [7,35]. On this basis, the length of KV primary cilia was measured in control and ptx3b morphants after acetylated α-tubulin immunostaining [6]. As shown in Figure 3, primary cilia of the KV are significantly shorter in ptx3b morphants when compared to control animals. At variance, no difference in the number of KV primary cilia was observed in ptx3b MO-injected embryos versus controls (mean ± S.E.M. equal to 21.5 ± 2.7 (n = 30) versus 22.7 ± 2.0 (n = 35) cilia per KV, respectively). To confirm the specificity of the ptx3b MO effects, one-cell stage embryos were co-injected with the ptx3b MO and an excess of hPTX3 mRNA. As shown in Figure 3D,E, hPTX3 mRNA was able to rescue the shortening of KV primary cilia caused by the ptx3b MO. Accordingly, rescued embryos showed a reduced percentage of cardiac looping defects when compared to ptx3b morphants ( Figure 3F). Notably, in keeping with the alterations observed in the KV, ptx3b morphants showed also structural defects of the primary cilia of the pronephric ducts [36] ( Figure 3G-J).
Given the capacity of PTX3 to bind different members of the FGF family [29], these data raise the hypothesis that a modulation of the activity of the FGF/FGFR system might be responsible for the shortening of primary cilia in ptx3b morphants. Indeed, ptx3b MO injection caused a significant increase of FGFR1 phosphorylation in zebrafish embryos that was fully abolished by the co-injection with an excess of hPTX3 mRNA ( Figure 4A,B). To substantiate this hypothesis, ptx3b MO-injected embryos were treated at the shield stage with the selective tyrosine kinase FGFR inhibitor BGJ398 (50 and of embryos injected with ctrl MO, ptx3b MO or ptx3b MO plus hPTX3 mRNA (500 pg/embryo). Each dot represents the mean length of primary cilia measured in a single KV. n, total number of examined KVs. Data were obtained in three independent experiments, *** p < 0.001, Student's t-test; (F) percentage of embryos showing cmcl 2+ normal or altered cardiac looping at 48 hpf after injection with ptx3b MO or ptx3b MO plus hPTX3 mRNA. Data are the mean of three independent experiments. * p < 0.05, ** p < 0.01, ANOVA. n, total number of examined embryos; (G-I) representative pictures of acetylated α-tubulin + ciliary axonemes in pronephric ducts of WT, ctrl MO, and ptx3b MO-injected embryos at 48 hpf stage. Scale bar: 10 µm; (J) percentage of embryos with normal or altered primary cilia organization in pronephric ducts. ** p < 0.01, Z-test, two-tailed.
Notably, in keeping with the alterations observed in the KV, ptx3b morphants showed also structural defects of the primary cilia of the pronephric ducts [36] ( Figure 3G-J).
Given the capacity of PTX3 to bind different members of the FGF family [29], these data raise the hypothesis that a modulation of the activity of the FGF/FGFR system might be responsible for the shortening of primary cilia in ptx3b morphants. Indeed, ptx3b MO injection caused a significant increase of FGFR1 phosphorylation in zebrafish embryos that was fully abolished by the co-injection with an excess of hPTX3 mRNA ( Figure 4A,B). To substantiate this hypothesis, ptx3b MO-injected embryos were treated at the shield stage with the selective tyrosine kinase FGFR inhibitor BGJ398 (50 nM in fish water). As anticipated, BGJ398 treatment was able to rescue the length of KV primary cilia in ptx3b morphants ( Figure 4C-F).
Cancers 2020, 12, 1756 6 of 17 nM in fish water). As anticipated, BGJ398 treatment was able to rescue the length of KV primary cilia in ptx3b morphants ( Figure 4C-F). Together, these data show for the first time that ptx3b regulates the length of primary cilium axoneme during zebrafish embryo development by modulating the activity of the FGF/FGFR system, thus playing a non-redundant role in the left-right asymmetry process and visceral organ positioning in zebrafish.

PTX3 Regulates Primary Cilium Length in Cancer Cells via the FGF/FGFR System
Alterations of primary cilia may contribute to cancer progression [11,12]. Based on the results obtained in ptx3b zebrafish morphants and the well-known impact of PTX3 on FGF-dependent tumors [23][24][25][26][27]37], we decided to investigate whether the modulation of PTX3 expression may affect primary cilia in FGF-dependent cancer cells.
Stemming from the observation that primary cilium number and length decrease in a subset of pre-invasive human prostatic lesions [13], we measured the length of primary cilia in TRAMP-C2 prostate cancer cells and in their hPTX3-overexpressing counterpart (hPTX3-TRAMP-C2 cells), characterized by a reduced FGFR signaling and tumorigenic potential [37]. As anticipated, hPTX3 overexpression resulted in a significant elongation of primary cilium in hPTX3-TRAMP-C2 cells when compared to TRAMP-C2 cells (Figures 5 and S4). Similar results were obtained in hPTX3 transfectants that originated from FGF-dependent human bladder cancer 5637 cells [27] and murine fibrosarcoma MC17-51 cells [26] (Figure 5). At variance, hPTX3 overexpression did not affect the percentage of ciliated tumor cells in all the cell lines tested ( Figure S5). Together, these data show for the first time that ptx3b regulates the length of primary cilium axoneme during zebrafish embryo development by modulating the activity of the FGF/FGFR system, thus playing a non-redundant role in the left-right asymmetry process and visceral organ positioning in zebrafish.

PTX3 Regulates Primary Cilium Length in Cancer Cells via the FGF/FGFR System
Alterations of primary cilia may contribute to cancer progression [11,12]. Based on the results obtained in ptx3b zebrafish morphants and the well-known impact of PTX3 on FGF-dependent tumors [23][24][25][26][27]37], we decided to investigate whether the modulation of PTX3 expression may affect primary cilia in FGF-dependent cancer cells.
Stemming from the observation that primary cilium number and length decrease in a subset of pre-invasive human prostatic lesions [13], we measured the length of primary cilia in TRAMP-C2 prostate cancer cells and in their hPTX3-overexpressing counterpart (hPTX3-TRAMP-C2 cells), characterized by a reduced FGFR signaling and tumorigenic potential [37]. As anticipated, hPTX3 overexpression resulted in a significant elongation of primary cilium in hPTX3-TRAMP-C2 cells when compared to TRAMP-C2 cells ( Figure 5 and Figure S4). Similar results were obtained in hPTX3 transfectants that originated from FGF-dependent human bladder cancer 5637 cells [27] and murine fibrosarcoma MC17-51 cells [26] (Figure 5). At variance, hPTX3 overexpression did not affect the percentage of ciliated tumor cells in all the cell lines tested ( Figure S5). PTX3 binds FGFs via its N-terminal domain [20]. To confirm the hypothesis that PTX3 may affect primary cilium length via the FGF/FGFR system in cancer cells, we measured the length of primary cilium in TRAMP-C2 cells stably transfected with the N-terminal hPTX3 cDNA (N-term-hPTX3-TRAMP-C2 cells) or with the C-terminal hPTX3 cDNA (C-term-hPTX3-TRAMP-C2 cells) [37]. As anticipated, overexpression of the FGF-binding N-terminal fragment of PTX3 resulted in an increase of the length of primary cilium when compared to mock-transfected cells, whereas overexpression of the PTX3 C-terminus was ineffective ( Figure 6A). These results were confirmed by a rescue experiment in which treatment with exogenous recombinant FGF2 protein caused the shortening of the primary cilium in N-term-hPTX3-TRAMP-C2 cells with no effect on mock-TRAMP-C2 cells ( Figure 6B).
In keeping with these observations, treatment with recombinant PTX3 protein or with the neutralizing anti-FGFR1 single-chain antibody fragment scFv-RR-C2 [38] resulted in the elongation of primary cilium in TRAMP-C2 cells ( Figure 6C). Accordingly, treatment with the tyrosine kinase FGFR inhibitors PD173074, SU5402 or BGJ398, or with different inhibitors of FGFR downstream signaling pathways, including the MAPK inhibitors PD98059 and U0126 and the PI3K inhibitors LY294002 and perifosine, were able to extend the length of primary cilium in TRAMP-C2 cells (Figure PTX3 binds FGFs via its N-terminal domain [20]. To confirm the hypothesis that PTX3 may affect primary cilium length via the FGF/FGFR system in cancer cells, we measured the length of primary cilium in TRAMP-C2 cells stably transfected with the N-terminal hPTX3 cDNA (N-term-hPTX3-TRAMP-C2 cells) or with the C-terminal hPTX3 cDNA (C-term-hPTX3-TRAMP-C2 cells) [37]. As anticipated, overexpression of the FGF-binding N-terminal fragment of PTX3 resulted in an increase of the length of primary cilium when compared to mock-transfected cells, whereas overexpression of the PTX3 C-terminus was ineffective ( Figure 6A). These results were confirmed by a rescue experiment in which treatment with exogenous recombinant FGF2 protein caused the shortening of the primary cilium in N-term-hPTX3-TRAMP-C2 cells with no effect on mock-TRAMP-C2 cells ( Figure 6B). Activation of different tyrosine kinase receptors suppresses ciliogenesis in retinal epithelial cells by stabilizing the trichoplein-Aurora A complex following phosphorylation of the deubiquitinase USP8 [39]. Accordingly, the elongation of primary cilium observed in hPTX3-TRAMP-C2 cells was paralleled by a significant reduction of the intracellular levels of trichoplein ( Figure 7A,B). Similar results were obtained in TRAMP-C2 cells treated with the tyrosine kinase FGFR inhibitors BGJ398 and SU5402 ( Figure 7C,D).
Sustained activation of the FGF/FGFR system may affect primary cilium-dependent Hh signaling following the shortening of primary cilia [30]. On this basis, we investigated the impact of PTX3 on the activation of the Hh pathway in TRAMP-C2 cells using the downregulation of the expression of the Hh target transcriptional regulator GLI1 as a readout [40,41]. As shown in Figure  7E, hPTX3 overexpression results in a significant increase of the levels of the Gli1 transcript in hPTX3-TRAMP-C2 cells that was abolished by stimulation with exogenous FGF2 ( Figure 7F). In keeping with these observations, treatment with recombinant PTX3 protein or with the neutralizing anti-FGFR1 single-chain antibody fragment scFv-RR-C2 [38] resulted in the elongation of primary cilium in TRAMP-C2 cells ( Figure 6C). Accordingly, treatment with the tyrosine kinase FGFR inhibitors PD173074, SU5402 or BGJ398, or with different inhibitors of FGFR downstream signaling pathways, including the MAPK inhibitors PD98059 and U0126 and the PI3K inhibitors LY294002 and perifosine, were able to extend the length of primary cilium in TRAMP-C2 cells ( Figure 6D,E). Together, the results indicate that the blockade of the FGF/FGFR system by extracellular or intracellular inhibitors exerts a significant impact on the primary cilium in TRAMP-C2 cells.
Activation of different tyrosine kinase receptors suppresses ciliogenesis in retinal epithelial cells by stabilizing the trichoplein-Aurora A complex following phosphorylation of the deubiquitinase USP8 [39]. Accordingly, the elongation of primary cilium observed in hPTX3-TRAMP-C2 cells was paralleled by a significant reduction of the intracellular levels of trichoplein ( Figure 7A,B). Similar results were obtained in TRAMP-C2 cells treated with the tyrosine kinase FGFR inhibitors BGJ398 and SU5402 ( Figure 7C,D).

The PTX3-Derived FGF Trap NSC12 Modulates Primary Cilium Length in Cancer Cells
Previous observations from our laboratory led to the discovery of NSC12 as the first PTX3derived small molecule endowed with a potent anti-FGF activity [24,42]. Orally available, NSC12 inhibits the growth, angiogenic potential, and metastatic activity of various FGF-dependent tumors with potential implications in cancer therapy [24]. On this basis, we evaluated the effect of NSC12 on cell proliferation and primary cilium length in FGF-dependent TRAMP-C2, 5637, and MC17-51 cells. As shown in Figures 8A and S6, NSC12 inhibits the proliferation of FGF-dependent tumor cells with an ID50 equal approximately to 1.0 μM. Accordingly, when administered at 1.0 μM concentration, NSC12 causes a significant elongation of the primary cilium in all the cell lines tested ( Figure 8B). Sustained activation of the FGF/FGFR system may affect primary cilium-dependent Hh signaling following the shortening of primary cilia [30]. On this basis, we investigated the impact of PTX3 on the activation of the Hh pathway in TRAMP-C2 cells using the downregulation of the expression of the Hh target transcriptional regulator GLI1 as a readout [40,41]. As shown in Figure 7E, hPTX3 overexpression results in a significant increase of the levels of the Gli1 transcript in hPTX3-TRAMP-C2 cells that was abolished by stimulation with exogenous FGF2 ( Figure 7F).

The PTX3-Derived FGF Trap NSC12 Modulates Primary Cilium Length in Cancer Cells
Previous observations from our laboratory led to the discovery of NSC12 as the first PTX3-derived small molecule endowed with a potent anti-FGF activity [24,42]. Orally available, NSC12 inhibits the growth, angiogenic potential, and metastatic activity of various FGF-dependent tumors with potential implications in cancer therapy [24]. On this basis, we evaluated the effect of NSC12 on cell proliferation and primary cilium length in FGF-dependent TRAMP-C2, 5637, and MC17-51 cells. As shown in Figure 8A and Figure S6, NSC12 inhibits the proliferation of FGF-dependent tumor cells with an ID 50 equal approximately to 1.0 µM. Accordingly, when administered at 1.0 µM concentration, NSC12 causes a significant elongation of the primary cilium in all the cell lines tested ( Figure 8B). Next, the capacity of NSC12 to exert a ciliogenic effect was evaluated in vivo on TRAMP-C2 tumor grafts. To this purpose, TRAMP-C2 cells were injected subcutaneously (s.c.) in syngeneic mice. At tumor take (30 days post-implantation), animals were treated i.p three times a week with NSC12 (5.0 mg/kg body wt) or vehicle. After 17 days, tumors were measured with calipers and harvested. Then, OCT embedded samples were immunostained with anti-acetylated α-tubulin antibody. In keeping with previous observations [24], NSC12 causes a significant inhibition of TRAMP-C2 tumor growth that was paralleled by a significant increase of the length of acetylated α-tubulin + tumor cilium (Figure 9). At variance, NSC12 did not cause any change in α-tubulin immunoreactivity in TRAMP-C2 grafts ( Figure S7).  Next, the capacity of NSC12 to exert a ciliogenic effect was evaluated in vivo on TRAMP-C2 tumor grafts. To this purpose, TRAMP-C2 cells were injected subcutaneously (s.c.) in syngeneic mice. At tumor take (30 days post-implantation), animals were treated i.p three times a week with NSC12 (5.0 mg/kg body wt) or vehicle. After 17 days, tumors were measured with calipers and harvested. Then, OCT embedded samples were immunostained with anti-acetylated α-tubulin antibody. In keeping with previous observations [24], NSC12 causes a significant inhibition of TRAMP-C2 tumor growth that was paralleled by a significant increase of the length of acetylated α-tubulin + tumor cilium (Figure 9). At variance, NSC12 did not cause any change in α-tubulin immunoreactivity in TRAMP-C2 grafts ( Figure S7). Next, the capacity of NSC12 to exert a ciliogenic effect was evaluated in vivo on TRAMP-C2 tumor grafts. To this purpose, TRAMP-C2 cells were injected subcutaneously (s.c.) in syngeneic mice. At tumor take (30 days post-implantation), animals were treated i.p three times a week with NSC12 (5.0 mg/kg body wt) or vehicle. After 17 days, tumors were measured with calipers and harvested. Then, OCT embedded samples were immunostained with anti-acetylated α-tubulin antibody. In keeping with previous observations [24], NSC12 causes a significant inhibition of TRAMP-C2 tumor growth that was paralleled by a significant increase of the length of acetylated α-tubulin + tumor cilium (Figure 9). At variance, NSC12 did not cause any change in α-tubulin immunoreactivity in TRAMP-C2 grafts ( Figure S7).

Discussion
The primary cilium is a microtubule-based structure that protrudes from the surface of most mammalian cells. It functions as a cellular antenna that captures environmental signals and serves as a hub for developmental and homeostatic signaling pathways [43].
PTX3 is a soluble patter recognition receptor and a key player of the innate immunity arm with non-redundant functions in pathogen recognition and inflammatory responses [15,16]. In addition, PTX3 is endowed with the capacity to exert an antitumor activity by acting as a natural FGF trap [29]. In this research, we provide experimental evidence that ptx3b, an ortholog of hPTX3, affects zebrafish development by regulating the length of the axoneme of the primary cilia of the KV, playing a non-redundant role in the left-right asymmetry process and visceral organ positioning during zebrafish embryogenesis. These effects appear to be mediated by the capacity of Ptx3b to modulate FGF/FGFR signaling in zebrafish embryo, its downregulation leading to increased FGFR1 phosphorylation. This was paralleled by shortening of the primary cilium axoneme in the KV, rescued by treatment with the selective tyrosine kinase FGFR inhibitor BGJ398. These data extend recent observations about the capacity of FGFR1 or FGF ligand inactivation to cause primary cilium shortening in zebrafish and Xenopus embryos [31]. Accordingly, FGF signaling regulates the length of primary cilia in various stem, embryonic, and differentiated mammalian cell types [30]. In addition, human skeletal dysplasia caused by activating FGFR3 mutations are characterized by alterations of the primary cilium [30,44].
Compelling experimental evidence indicates that alterations of primary cilium may play an important role also in tumors by affecting various intracellular signal transduction pathways (reviewed in [8,12]). Moreover, alterations of primary cilium affect different aspects of tumor biology, including cancer cell autophagy and apoptosis, response to hypoxia, epithelial-mesenchymal transition, and drug resistance [45]. In this frame, various types of cancer cells fail to express the primary cilium, including pancreatic, breast, prostate cancer, and melanoma cells [13,[45][46][47][48][49][50]. Hence, restoration of the primary cilium in cancer cells may provide novel opportunities for therapeutic antineoplastic interventions. In this frame, drug repurposing "ciliotherapy" approaches have been proposed to hamper tumor growth by induction of the primary cilium [14].
Deregulation of the FGF/FGFR network occurs in tumors due to gene amplification, activating mutations and oncogenic fusions [51][52][53]. The multifaceted FGFR signaling network is engaged in the progression of different FGF-dependent tumors by acting on both tumor and stromal cell compartments, thus affecting oncogenesis through different mechanisms, including cell-signaling deregulation, angiogenesis, and resistance to cancer therapies [54]. On this basis, different classes of therapeutics have been developed, including non-selective and selective tyrosine kinase FGFR inhibitors, monoclonal antibodies, and FGF ligand traps [55,56]. Some of them are under investigation in clinical trials in different FGFR-related cancer settings [53].
In this frame, PTX3 has been shown to act as an oncosuppressor in different FGF-dependent tumors [24] by inhibiting tumor growth, neovascularization, and metastatic dissemination [29]. Here, we demonstrate that PTX3 has a ciliogenic effect on different cancer cell types. Experimental evidences indicate that this capacity is due to its FGF trap activity. (i) Overexpression in TRAMP-C2 cells of the FGF binding N-terminal fragment of PTX3 resulted in the elongation of primary cilium, whereas overexpression of the PTX3 C-terminus was ineffective; (ii) tyrosine kinase FGFR inhibitors and inhibitors of FGFR downstream signaling pathways cause primary cilium elongation in TRAMP-C2 cells. In parallel, tyrosine kinase FGFR inhibitors cause a significant reduction of the intracellular levels of trichoplein that, in complex with Aurora A, exerts suppressive effects on ciliogenesis [39]; (iii) hPTX3 overexpression leads to Gli1 upregulation, a marker of cilium-dependent Hh signaling [40], which is prevented by treatment with exogenous FGF2 protein.
NSC12 is a PTX3-derived small molecule able to bind all the extracellular members of the FGF family [24]. Numerous experimental evidences indicate that NSC12 may represent a prototype for the development of novel orally available therapeutic agents targeting FGF-dependent tumors [53]. Indeed, when compared to tyrosine kinase FGFR inhibitors, NSC12 is characterized by a reduced toxicity that may result in a more favorable therapeutic window [54]. Here we show that NSC12 inhibits the proliferation of FGF-dependent murine prostate cancer TRAMP-C2 cells, human bladder carcinoma 5637 cells, and murine fibrosarcoma MC17-51 cells (see also [26,27,37]) and that this inhibition is paralleled by the elongation of the primary cilium in all the cell lines tested. These data were confirmed in vivo, where NSC12 was able to inhibit tumor growth, and to induce a ciliogenic effect in TRAMP-C2 tumor grafts.
The present research focused mainly on the effect on primary cilium of FGF2 and FGFR1, the prototypic members of the FGF/FGFR system. Thus, further experiments will be required to assess the impact exerted on primary cilium by the modulation of different FGFs and FGFRs. Nevertheless, given the capacity of NSC12 to inhibit the activity of all the canonical FGFs and their interaction with all FGFRs [24], this small molecule FGF trap may represent the basis for the design of novel ciliogenic anticancer drugs able to affect different aspect of the biology of FGF-dependent tumors.

Zebrafish Maintenance and Collection
The wild-type zebrafish AB line was maintained at 28 • C under standard conditions and embryos were staged as described [57]. To examine embryos older than 22 hpf, fish water was added with 0.2 mM 1-phenil-2-thiourea (PTU, Sigma-Aldrich). For the observation of the in vivo phenotypes, embryos were anesthetized using 0.16 mg/mL Tricaine (Sigma).
A five-mismatch nucleotide ctrl MO (5 -CTCAATGATCTACCTCACGGCAGAT-3 ) was used as control. To confirm the targeting efficacy of the ptx3b MO, alternative splicing pattern analysis was performed on zebrafish embryos using appropriate primers (Table 1). Table 1. Oligonucleotide primers used for RT-PCR and qPCR analysis.

Gene
Forward Reverse

Whole-Mount In Situ Hybridization
Digoxigenin-labelled RNA probes were transcribed from linear cDNA constructs (Roche Applied Science, Penzberg, Germany). WISH was performed on embryos fixed in 4% PFA as described [58]. For sectioning, zebrafish embryos were post-fixed in 4% PFA after WISH, dehydrated in ethanol series, cleared in xilol, and paraffin-embedded overnight.

RT-PCR and qPCR Analysis
Total RNA was isolated from untreated and MO-injected embryos at different stages of development using TRIzol ® Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Two micrograms of total RNA were retrotranscribed and 100 ng of cDNA were used for the evaluation of the alternative ptx3b splicing pattern by semi-quantitative RT-PCR analysis. The whole gels are shown in Figure S4.
For the evaluation of Gli1 expression in cancer cells, total RNA was isolated with the same procedure. Then, 1/10th of the retrotranscribed cDNA was used for quantitative qPCR that was performed with the ViiA 7 Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA) using iTaq Syber Green Supermix (Biorad, Hercules, CA, USA) according to the manufacturer's instructions. Samples were analyzed in triplicate and normalized with respect to the levels of Gapdh expression using the appropriate primers (Table 1).

Cell Proliferation Assay
Cells were seeded on 48-well plates at 8 × 10 3 cells/cm 2 . After 24 h, cells were treated with increasing concentrations of NSC12 in serum free DMEM (TRAMP-C2 and MC17-51 cells) or RPMI (5637 cells). After a further 48 h incubation, cells were trypsinized and viable cell counting was performed with the MACSQuant ® Analyzer (Miltenyi Biotec, Bergisch Gladbach, Germany).

Tumor Grafts in Mice
Animal studies were approved by the local animal ethics committee (Organismo Preposto al Benessere degli Animali, Università degli Studi di Brescia, Brescia, Italy) and by the Italian Ministero della Salute. All the procedures and animal care were conformed to institutional guidelines that comply with national and international laws and policies (EEC Council Directive 86/609, OJL 358, 12 December 1987). C57BL/6 (Charles River, Calco, Italy) male mice were maintained under standard housing conditions. TRAMP-C2 cells (5 × 10 6 in 200 µL of PBS) were injected s.c. into the dorsolateral flank of nine week-old animals. At tumor take (30 days post-implantation), animals were treated i.p three times a week with NSC12 (5.0 mg/kg body wt) or vehicle. Tumors were measured in two dimensions and volume was calculated according to the formula V = (D × d 2 )/2, where D and d are the major and minor perpendicular tumor diameters, respectively. After 17 days, tumors were harvested, embedded in OCT compound, and immediately frozen.

Immunoflurescence Analysis
Control and ptx3b MO injected embryos were fixed in 4% paraformaldehyde for 2 h at room temperature. Whole-mount immunofluorescence analysis of cilium axoneme was performed using a mouse anti-acetylated α-tubulin antibody (1:1000, Sigma) followed by Alexa Fluor 488 anti-mouse IgG (Molecular Probes, Eugene, OR, USA).
For tumor samples, 15 µm cryostat sections were air dried, fixed with 4% paraformaldehyde for 15 min, and permeabilized with 0.5% Triton X-100/PBS for 15 min at room temperature (RT). Sections were blocked for 2 h at RT with 1% BSA, 0.1% fish gelatin (Sigma-Aldrich), 0.1% Triton X-100 and 0.05% Tween20 (blocking solution), and incubated for a further 1 h at RT with a M.O.M blocking reagent (Vector Laboratories, Burlingame, CA, USA) to block endogenous mouse IgGs. Then, sections were incubated overnight at 4 • C with anti-acetylated α-tubulin antibody (1:1000 in blocking solution) and for one further hour at RT with anti-mouse Alexa Fluor 488 secondary antibody (1:250) in blocking solution. Nuclei were counterstained with DAPI.
Immunostained cells, embryos, and tissue sections were analyzed using a Zeiss Axiovert 200M epifluorescence microscope equipped with a Plan-Apochromat × 63/1.4 NA oil objective. Cilium length (in µm) was quantified manually using the "segmented line" tool of ImageJ (NIH).

Conclusions
In summary, our study demonstrates the ability of the natural FGF trap PTX3 to exert a modulatory effect on primary cilium in embryonic development and cancer. Moreover, they set the basis for the design of novel PTX3-derived ciliogenic drugs with potential implications for the therapy of FGF-dependent tumors.