A ligand-insensitive UNC5B splicing isoform regulates angiogenesis by promoting apoptosis

Abstract

The Netrin-1 receptor UNC5B is an axon guidance regulator that is also expressed in endothelial cells (ECs), where it finely controls developmental and tumor angiogenesis. In the absence of Netrin-1, UNC5B induces apoptosis that is blocked upon Netrin-1 binding. Here, we identify an UNC5B splicing isoform (called UNC5B-Δ8) expressed exclusively by ECs and generated through exon skipping by NOVA2, an alternative splicing factor regulating vascular development. We show that UNC5B-Δ8 is a constitutively pro-apoptotic splicing isoform insensitive to Netrin-1 and required for specific blood vessel development in an apoptosis-dependent manner. Like NOVA2, UNC5B-Δ8 is aberrantly expressed in colon cancer vasculature where its expression correlates with tumor angiogenesis and poor patient outcome. Collectively, our data identify a mechanism controlling UNC5B's necessary apoptotic function in ECs and suggest that the NOVA2/UNC5B circuit represents a post-transcriptional pathway regulating angiogenesis.

Fig. 1. Inclusion of UNC5B exon 8 is regulated in ECs.

a Differentially spliced exonic events in NOVA2-depleted moEC compared to control moEC (NCBI BioProject: PRJNA293346). Absolute difference of Percent Spliced-In (PSI) between means is shown. Unc5b exon 8 AS event is indicated in red. b Schematic representation of the mouse and human UNC5B genomic region with the AS exon 8 of 33 nt (gray). Black boxes = constitutive exons; lines = introns. Two different mRNAs result from the inclusion (red lines) or skipping (blue lines) of exon 8. Arrows indicate primers used for RT-PCR. c Left: RT-qPCR (relative to Ubb) and immunoblotting of NOVA2 expression levels in NOVA2-depleted moEC. The anti-NOVA2-specific antibody recognizes two immunoreactive bands at 50–55 kDa and 70–80 kDa, as previously reported^, ; VINCULIN as loading control. Right: RT-PCR of Unc5b exon 8 AS profile in the above ECs. n = 3. d Nova2 mRNA (relative to mouse Ubb or human ACTB) and protein expression levels, and AS analysis of Unc5b exon 8 in another murine EC line (luEC) (n = 3) and e in human immortalized ECs (HUVEC/TERT2) transfected with a siRNA against Nova2 or with a control siRNA. VINCULIN as loading control. n = 5. f Left: in vitro angiogenesis assay of HUVEC/TERT2 plated on Matrigel-coated plates. Scale bar: 0.5 mm. Right: NOVA2 mRNA expression levels (relative to B2M) and UNC5B exon 8 AS during differentiation of HUVEC/TERT2 and formation of capillary tube-like structures on Matrigel. h = hours after seeding. n = 3. g Analysis, by RT-qPCR, of Nova2 mRNA expression levels (relative to Ubb) and, by RT-PCR, of Unc5b exon 8 splicing in freshly purified ECs from mouse lung and colon tissue. n = 3. h Nova2 mRNA expression levels (relative to Ubb) and Unc5b exon 8 AS in primary ECs from mouse hearth (cardiac ECs) at different developmental stages: embryonic (E16–E18); neonatal (P0-P1); adult (P60). n = 3. The percentage of exon inclusion is shown under each gel. n = biologically independent experiments. Two-tailed Student’s t-test or one-way ANOVA for multiple comparisons; Error bars indicate ±SEM. Exact P values are indicated: *P < 0.05; **P < 0.01; ***P < 0.001; ns not significant.

Fig. 2. NOVA2 directly binds to Unc5b pre-mRNA.

a Schematic representation of mouse and human UNC5B pre-mRNAs. AS exon 8 in gray. Constitutive exons 7 and 9 in black. Introns 7 and 8 are also indicated (thin lines). YCAY sites, identified by using the RBPmap tool within 200 nt from the exon–intron junctions, are indicated in orange, pale orange, or yellow bars depending on their Z-score and P values (calculated by RBPmap tool). b Upper: schematic representation of mouse Unc5b as in a. Arrows indicate primers used for RT-qPCR of NOVA2 immunoprecipitated RNAs. Lower: RT-qPCR of NOVA2 CLIP experiments in moEC. Amplification of immunoprecipitated RNAs with primers annealing to the YCAY cluster in intron 7 (A) and, as negative controls, within intron 8 (B and C). Error bars indicate ±SD calculated from n = 3 independent experiments. c Upper: the mouse Unc5b genomic region encompassing exons 7, 8, and 9, cloned into the pcDNA3.1(+) vector to generate the p-Unc5b wild-type (Wt) minigene. Arrows indicate primers used for splicing analysis of the transcripts generated from the minigene. Boxes = exons; thin lines = introns; pCMV = promoter; BGHpA = polyadenylation sequence. Colored vertical bars as in a. Lower: p-Unc5b Wt co-transfected in moEC with either NOVA2-HA (NOVA2), T7-HNRNPA1 (HNRNPA1), or empty vector (Vector). RT-PCR analysis of p-Unc5b splicing is shown. Ectopic expression of NOVA2 and HNRNPA1 were confirmed by immunoblotting with anti-HA and anti-T7 antibodies, respectively. VINCULIN as loading control. d Splicing assay with p-Unc5b Mut in moEC. Red asterisks indicate the mutated NOVA2 binding sites (YCAY repeats were replaced with YAAY). e NOVA2 (yellow circle) promotes UNC5B exon 8 skipping by directly binding to the YCAY cluster (orange bar) located in the intronic region upstream exon 8. The percentage of exon inclusion is shown below each gel. At least three independent biological replicates were analyzed for each experiment.

Fig. 3. Unc5b-Δ8 promotes PAV vessel formation in zebrafish.

a Schematic representation of mouse, human, and zebrafish unc5b genomic regions with the AS exon 8 in gray. Black boxes = constitutive exons; lines = introns. YCAY motifs identified with RBPmap tool are indicated as vertical bars. Arrows indicate primers for RT-PCR. b Splicing analysis of unc5b exon 8 and nova2 mRNA expression levels in zebrafish embryos at different developmental stages: maternal, 24 h post-fertilization (hpf), and 48 hpf. c RT-PCR with RNA extracted from zebrafish embryos (28 hpf) injected with a control morpholino (MO-ctr) or a morpholino against nova2 (MO-nova2). Altered AS was partially corrected by the co-injection of a morpholino-resistant nova2 mRNA (MO-nova2+nova2). n = 6. d Compared to the wild-type (wt) organism, nova2 mutants show a reduction of the unc5b-Δ8 transcript (72 hpf). The percentage of exon inclusion is indicated. n = 3. e Left: scheme of zebrafish vessels; PAV parachordal vessel, DA dorsal aorta, PCV posterior cardinal vein, ISV intersegmental vessel. Central: lateral views (fluorescence) of Tg(kdrl:GFP) ^la116 zebrafish embryos (72 hpf) injected with a control morpholino (MO-ctr) or with a morpholino against unc5b (MO-unc5b); unc5b morphants were also co-injected with morpholino-resistant zebrafish mRNAs encoding for unc5b AS isoforms (unc5b-FL or unc5b-Δ8). Right: quantification of embryos in which PAV is correctly formed (%). n = 3. Arrows indicate correctly forming PAV, whereas arrowheads indicate those unproperly forming. f Left: Lateral views (fluorescence) of 72 hpf wild-type (wt) and nova2 CRISPR mutant zebrafish embryos. Right: Quantification of PAV formation in 72 hpf wild-type (wt) and nova2 mutant (mut). N. of larvae with normal PAV out the total n. of larvae. Arrowheads indicate forming PAV. g Left: lateral views (fluorescence) of wt and nova2 mutant embryos (6 dpf) injected with mRNAs encoding for unc5b AS isoforms (unc5b-FL or unc5b-Δ8). Arrows indicate forming PAV. Right: quantification of embryos in which PAV is correctly formed (%). n = 3. n = biologically independent experiments. One-way ANOVA for multiple comparisons; Error bars indicate ±SEM. Exact P values are indicated: *P < 0.05; ***P < 0.001; ****P < 0.0001. Scale bars: 50 μm in e, 100 μm in f, 25 μm in g.

Fig. 4. UNC5B-Δ8 prevents Netrin-1 prosurvival signaling and promotes vessel formation in an apoptotic-dependent manner.

a Schematic representation of Netrin-1/UNC5B-mediated prosurvival signal. UNC5B domains are indicated with different colors. Blue ovals: Ig-like domains; red ovals: thrombospondin-like domains (TSP1); beige cylinder: transmembrane domain (TM); yellow circle: ZU5 domain; violet cube: UPA/DCC-binding domain (DB); pink circle: death domain (DD). Region encoded by exon 8 is shown in green. Netrin-1 is in orange, whereas DAPk is in dark blue. DAPk phosphorylation at Ser308 is indicated. b Localization of GFP-tagged UNC5B isoforms in HUVECs. Scale bar: 10 μm. c Co-immunoprecipitation of UNC5B AS isoforms GFP-tagged with HA-tagged isoforms in the presence or absence of Netrin-1 (150 ng/ml). Unspecific magnetic beads (Beads) as control. d Vital exclusion assay in HeLa cells overexpressing UNC5B AS isoforms or GFP as control. Error bars indicate ±SD; n = 3. One-way ANOVA for multiple comparisons. e DAPk phosphorylation (Ser308) was evaluated in HeLa cells co-transfected with plasmids expressing DAPk and Unc5b isoforms with or without Netrin-1 treatment (150 ng/ml). Total DAPk as control. f Co-immunoprecipitation of Phospho-DAPk (Ser308) in HeLa cells transiently transfected as in e. g Left: Lateral views (fluorescence) of 72 hpf zebrafish embryos expressing the GFP under the control of the endothelial-specific promoter kdrl injected with a control morpholino (MO-ctr) or with a morpholino against unc5b (MO-unc5b); unc5b morphants were also co-injected with a morpholino-resistant zebrafish mRNA encoding for unc5-Δ8 (unc5b-Δ8) and treated with a pan caspase inhibitor (BAF inhibitor). White arrows: forming PAV. Right: Quantification of PAV formation in the above zebrafish embryos, two different biological replicates were analyzed. n = 2. Error bars indicate ±SEM. Scale bar: 50 μm. h HUVEC/TERT2 treated with a siRNA oligo against the UNC5B-Δ8 mRNA (n  5) or a control (Ctr) oligo (n = 6) were analyzed (left) by RT-PCR for UNC5B exon 8 splicing and (center) for the formation of capillary tube-like structures on Matrigel. Right: quantification of nodes and segments (# per field) in the in vitro tube formation assay. Two-tailed Student’s t-test. Error bars indicate ±SD. n = biologically independent experiments. Exact P values are indicated: *P < 0.05;; ***P < 0.001; ****P < 0.0001; ns not significant.

Fig. 5. NOVA2 and UNC5B-Δ8 expression levels in colon cancer patients.

a IHC of NOVA2 (I and III, scale bar: 60 μm) and ISH with the UNC5B-Δ8 probe (II and IV, scale bar: 30 μm) in colon primary tumors and normal adjacent tissues. Black arrowheads indicate UNC5B-Δ8 signal in ECs in tumor blood vessels. b RNA-seq data analysis of NOVA2 mRNA expression levels, c PSI (Percent Spliced-In) of UNC5B exon 8, and d correlation between NOVA2 and UNC5B-Δ8 expression in normal (blue) and colon adenocarcinoma (red) samples in the TCGA-COAD dataset. e NOVA2 and f UNC5B-Δ8 correlation with a colon tumor-specific angiogenetic signature (TEC). g NOVA2 mRNA Z-score expression levels in colon adenocarcinoma without (N0) or with metastasis (N1 or N2) annotated in the TCGA-COAD dataset. h NOVA2 mRNA expression levels in colon adenocarcinoma without or with metastasis (GSE14095 and GSE87211 datasets; http://hcmdb.i-sanger.com). i Kaplan–Meier plot of 5-years overall survival in colon adenocarcinoma patients classified according to NOVA2 or l UNC5B-Δ8 mRNA expression. Blue curve: high expression (1st quartile); red curve: moderate expression; green curve: low expression (3rd quartile). Log-ranked P-values are indicated. Boxplot represents upper quartile, median, lower quartile. Error bars indicate minimum and maximum. Two-tailed Log-ranked test (in b) or Two-tailed Student’s t-test (in c and g) and exact P values are indicated: *P < 0.05, ****P < 0.0001. P values in h, were calculated by HCMDB platform. For correlation analysis, linear regression (black line) and Pearson r coefficient with two-tailed P-value were also calculated.