Interleukin-30 subverts prostate cancer-endothelium crosstalk by fostering angiogenesis and activating immunoregulatory and oncogenic signaling pathways

For mouse studies, sample sizes were determined by minimizing the number of animals essential to statistically significant results (in accordance with the 3R principles). An overall sample size of 15 mice per group allowed the detection of a statistically significant difference between the experimental groups, with an 80% power, at a 0.05 significance level. Animals were randomly assigned to study arms.

For studies on human tissue samples, a cohort of 80 patients allowed the detection of a statistically significant correlation between two genes, with an 85% power and a 5% significance level (G*Power, RRID:SCR_013726). PC patients who had not received immunosuppressive treatments, hormone- or radiotherapy, and were free from immune system diseases, were selected by matching for Gleason score with patients from the Prostate Cancer Transcriptome Atlas (PCTA) (Table 1).

All experiments were performed in blind. Study investigators were unaware to which group a particular animal was assigned to, and if a particular human PC sample was IL30 negative or positive.

Primary human umbilical vein endothelial cells (HUVEC; #PCS-100–010) and immortalized human aortic endothelial cells (TeloHAEC; RRID:CVCL_Z065) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and were cultivated in Vascular Cell Basal Medium (#PCS-100–030; ATCC) plus Endothelial Cell Growth Kit-VEGF (#PCS-100–041; ATCC). Human PC cell lines, AR⁻ PC3 and AR⁺ DU145 [15, 19, 20], were purchased from the ATCC, which authenticated them by short tandem repeat profile analysis. PC cells were cultured in RPMI-1640 (#R8758; Merck, Darmstadt, Germany) with 10% fetal calf serum (#F1283; Merck).

For PC-Endothelial cell (EC) cocultures, PC and ECs, were seeded in 1:1 ratio and cultivated in Vascular Cell Basal Medium (ATCC; #PCS-100–030) plus Endothelial Cell Growth Kit-VEGF (ATCC; #PCS-100–041). Cell viability and proliferation were assessed using the CellTiter 96 AQueous One Solution Cell Proliferation Assay (#G3582; Promega, Madison, WI, USA), according to manufacturer’s instructions.

Real-time RT-PCR and PCR array were performed, as described in the Supplementary Methods, using the RT² Profiler Human Angiogenesis PCR Array (#PAHS-024ZR), RT² Profiler Human Cancer Inflammation & Immunity Crosstalk PCR Array (#PAHS-181Z) and the RT² Profiler™ Human Prostate Cancer PCR Array (#PAHS-135Z) (all from Qiagen, Hilden, Germany).

To investigate the effect of IL30, produced by tumor cells, on endothelial cell proliferation, excluding the effect of other cytokines or soluble factors present in the extracellular matrix, we used the Growth Factor Reduced Matrigel (GFR-Matrigel), a soluble basement membrane extract, which contains the same growth factors present in standard Matrigel but in reduced concentrations. Briefly, the GFR-Matrigel (#354,230; BD Biosciences, Franklin Lakes, NJ, USA) was thawed on ice, at + 4 °C overnight, and then plated (125 µl/well) in 8-well chamber slides, which were incubated, at 37 °C for 30 min, to allow the Matrigel to polymerize [21]. Subsequently, 2 × 10⁵ HUVECs or HAECs were mixed with 2 × 10⁵ IL30-overexpressing DU145 or PC3, IL30KO-DU145 or -PC3, or the respective control cell lines, and added to the top of the Matrigel in each well. The chambers were then incubated at 37 °C for 24 h, fixed with acetone and stained as described in the section Histology, immunohistochemistry and morphometric analyses.

Endothelial cell counts were assessed by confocal microscopy, using a LSM 800 confocal microscope (Zeiss, Oberkochen, Germany; RRID:SCR_015963) and the ZEN Microscopy Software (Zeiss; RRID:SCR_013672). Four to six high-power fields were analyzed for each well and two optical sections per well were evaluated. Results were expressed as mean ± SD of CD31⁺cells per field.

After thawing, GFR-Matrigel (BD Biosciences) was added (125 µl/well) into 8-well plates and allowed to polymerize at 37 °C for 30 min. Meanwhile, HUVEC or HAEC were detached from flasks and re-suspended in culture medium with or without rIL30 (50 ng/ml). Subsequently, 2 × 10⁵ HUVEC or HAEC were added to the top of the Matrigel in each well. The slides were then incubated at 37° for 24 h and the tube formation was evaluated, after CD31 immunofluorescence staining, by confocal microscopy, using the LSM 800 confocal microscope and the ZEN Microscopy Software (both from Zeiss). Capillaries were identified as small tubes or circles marked by CD31 staining. Four to six high-power fields were analyzed for each well and two optical sections per well were evaluated. Results were expressed as mean ± SD of CD31⁺capillaries/field.

To assess phenotype markers and investigate surface expression of cytokines, cytokine receptors, growth factor receptors and adhesion molecules, human ECs were harvested and mechanically dissociated into a single cell suspension. Then, the cells were pelleted, resuspended in PBS and incubated for 30 min, at 4 °C, with the antibodies (Abs) listed in the Supplementary Table S1. Acquisition was performed using a BD Scientific Canto II Flow Cytometer (RRID:SCR_018056; BD Biosciences) and the data were analyzed using FlowJo software (RRID:SCR_008520; BD Biosciences). Dead cells were excluded by 7AAD staining. All experiments were performed in triplicate. To isolate PC and ECs for molecular studies, after their coculture, fluorescence-activated cell sorting (FACS) analyses were performed, using a BD FACS Aria II Cell Sorter (RRID:SCR_018091; BD Biosciences), as described in the Supplementary Methods.

For the overexpression of human IL30 in DU145 and PC3 cells, we used the IL27p28 Human Tagged ORF Clone (#RC209337L1; Origene, Rockville, MD, USA) which was transfected in cancer cells using Lipofectamine 3000 Reagent (#L3000001; Thermo Fisher Scientific, Waltham, MA, USA) as we described in ref. 15. The expression of IL30 was confirmed by real-time RT-PCR and western blot (WB) [15].

The CRISPR/Cas9 technology was used to generate IL30 gene knockout (IL30KO) DU145 and PC3 cells, as we described in ref.15. IL30 gene knockout was validated by WB [15].

Quantitation of ANG, CXCL10, EDN1 and IGF1, in the supernatant derived from human endothelial cells, was carried out using the following ELISA kits, according to manufacturer’s instructions: Angiogenin Human ELISA Kit (#EHANG); CXCL10 Human ELISA Kit (#KAC2361); Endothelin-1 Human ELISA Kit (#EIAET1) (all from Thermo Fisher Scientific, Waltham, MA, USA) and Human IGF1 ELISA Kit (#ab211651, Abcam, Cambridge, UK).

WB was performed, as described in the Supplementary Methods, to assess IL30, EBI3, ANG, CXCL9, EDN1, TGFB2, CXCL6, THBS2 and IGF1 protein expression in human endothelial cells, and IL1β, IL4, IL6, EGF, VEGFA, LGALS4 and SHBG protein expression in human PC cells.

The Proteome Profiler Human Phospho-Kinase Array (#ARY003B; R&D Systems, Minneapolis, MN, USA) was used according to manufacturer’s instructions. Briefly, cells were lysed in manufacturer’s buffer, protein were quantified by Bradford Protein Assay (Bio-Rad, Hercules, CA, USA) and samples adjusted to 800 μg/ml with lysis buffer. Then, 334 μl of lysate was loaded per membrane and signals were detected by Chemi-Reagent Mix. The signal intensity of each spot was determined by ImageJ software (RRID:SCR_003070) and results were expressed as mean ± SD of pixel density. Reference spots were used to normalize signals across membranes. All experiments were performed in triplicate.

NSG mice (RRID:IMSR_JAX:005557) were purchased from Charles River Laboratories (Wilmington, MA, USA). NSG mice were housed under high barrier conditions, according to the Jackson Laboratory’s guidelines, in the animal facility of the Center for Advanced Studies and Technology, Chieti, Italy.

To study in vivo the effects of IL30 overexpression or knockout, in PC cells, on tumor vasculature and expression of genes driving angiogenesis, inflammation and prostatic carcinogenesis, three groups (15 mice per group) of 8-week-old NSG mice were subcutaneously injected with 3 × 10⁵ wild-type (CTRL), Empty Vector (EV) or hIL30 lentiviral-DNA (IL30LV-DNA) transfected DU145 or PC3 cells, and another three groups of fifteen 8-week-old NSG mice, with 5 × 10⁵ wild-type (CTRL), non-targeting guide RNA-treated (NTgRNA) or IL30 knockout (IL30KO) DU145 or PC3 cells. Tumors were measured with calipers as soon as they were palpable, and mice were sacrificed when the tumor reached 700 mm³, since at this size there are still no important necrotic phenomena that can invalidate the immunohistochemical examination. An overall sample size of 15 mice per group allowed the detection of a statistically significant difference between the three groups, with an 80% power, at a 0.05 significance level (G*Power, RRID:SCR_013726).

Animal procedures were performed in accordance with the European Community and ARRIVE guidelines and were approved by the Institutional Animal Care Committee of “G. d’Annunzio” University and by the Italian Ministry of Health (Authorization n. 892/2018-PR).

For bioinformatic analyses, we used the “Prostate Cancer Transcriptome Atlas (PCTA)”, the largest of the publicly available online databases, which includes RNA-Seq data from 1116 clinical PC specimens, with annotation of Gleason score, collected from 38 PC datasets and normalized by median centering method and quantile scaling (http://​www.​thepcta.​org and ref 47). We used the resulting merged database to measure the statistical correlation between pairs of genes or between single genes and the apoptotic signaling pathway dataset reported in the PCTA (Supplementary Table S2). All statistical analyses were performed by applying the Spearman's rank correlation coefficient (ρ) calculation tool included in the PCTA website, at an α level of 0.05.

Tissue samples were collected and stored in the institutional Biobank of the Local Health Authority n. 2 Lanciano Vasto Chieti (Italy) and the personal data processing complies with Data Protection Laws. For this study, we selected, by matching for Gleason score with patients from the PCTA (Table 1), a validation cohort of 80 patients, who underwent radical prostatectomy for PC and had not received immunosuppressive treatments, hormone- or radiotherapy, and were free from immune system diseases. This sample size allowed the detection of a statistically significant correlation between two genes, with an 85% power and a 5% significance level (G*Power, RRID:SCR_013726). The study was approved by the Ethical Committee of the “G. d’Annunzio” University and Local Health Authority of Chieti (PROT. 1945/09 COET of 14/07/2009, amended in 2012), and was performed, after written informed consent from patients, in accordance with the principles outlined in the Declaration of Helsinki.

For histology, tissue samples were fixed in 4% formalin, embedded in paraffin, sectioned at 4 μm and stained with hematoxylin and eosin. Immunofluorescent stainings for CD31 and EpCAM and immunostainings for CD31, Ki67, CXCR5, EpCAM, IL12β, IL30, IGF1, LGALS4, NOS2, SHBG, TGFα and TNFα, were performed as described in ref. 18, using the Abs listed in the Supplementary Table S3. Proliferation index, microvessel counts and expression of immunoregulatory and prostate cancer driver genes in tumor samples were assessed as described in the Supplementary Methods. The morphometric analysis, on single immunostained sections, was performed by light microscopy with a Leica Imaging Workstation and QWin image analysis software (Leica QWin, RRID:SCR_018940). The Spearman's rank correlation coefficients for each pair of markers were calculated using Stata V.13 (StataCorp, College Station, TX, USA; RRID:SCR_012763).

For in vitro studies, in vivo immunohistochemical analyses on tumor xenografts and on human PC samples, for which data are approximately normally distributed, between-group differences were assessed by Student’s t-test, or ANOVA followed by Tukey HSD test. Spearman's correlation coefficient (ρ) was used to analyse correlations between the expression of IL30 and that of molecules resulting from bioinformatic findings. For the bioinformatics, statistical analyses have been described above.

All statistical tests were two-sided and evaluated at an α level of 0.05 using Stata V.13 (StataCorp, College Station, TX, USA; RRID:SCR_012763).

The data generated in this study are available upon request from the corresponding author. Expression profile data analyzed in this study were obtained from the Prostate Cancer Transcriptome Atlas (PCTA) collection, at http://​www.​thepcta.​org.

Primarily found in high-grade and stage of PC, IL30 expression is associated with a thriving vasculature, immune suppression, and tumor progression [14‐16]. To assess its potential impact on the PC-EC crosstalk, we used two human cell lines derived from the metastases of high-grade PCs, and that express membrane-anchored IL30 [15], namely DU145 cells [19], endowed with a CK8/14⁺AR⁺PSA⁺ phenotype, and castration resistant PC3 cells [20], endowed with a CD44⁺AR⁻PSA⁻CgA⁺NSE⁺ neuroendocrine phenotype. Both cell lines were engineered to overexpress IL30, henceforth referred to as IL30-DU145 and IL30-PC3 cells or were knocked out (KO) for the IL30 gene by CRISPR/Cas9 genome editing, hereinafter referred to as IL30KO-DU145 and IL30KO-PC3 cells [15]. Since the tumor vascular bed stems from the activation and proliferation of normal pre-existing vessels, for the PC-EC coculture experiments, we used primary ECs derived from human umbilical vein, HUVEC, and immortalized ECs isolated from human aortic endothelia, TeloHAEC, hereinafter referred to as HAEC. Both cell types were authenticated by STR (Eurofins Genomics, Ebersberg, Germany) and surface staining for characteristic markers (Supplementary Fig. S1).

Flow cytometric and western blot analysis demonstrated that HUVEC and HAEC did not express or produce IL30 (IL27/p28) (Fig. 1A, and Supplementary Fig. S2), but expressed CD130 (IL6R-beta) and CD126 (IL6R-alpha), which are currently known to function as the IL30 receptor (R) chains [22] (Fig. 1A).

Coculture with wild type PC cells, DU145 and PC3, increased (Student's t-test: p < 0.01) the proliferation of both HUVEC and HAEC, whereas deletion of the IL30 gene in PC cells dramatically decreased (Student's t-test: p < 0.01) EC proliferation, which was substantially improved by coculture with IL30 overexpressing PC cells (Student's t-test: p < 0.01), as shown by the flow cytometric analyses of Ki67 staining (Fig. 1B). EC apoptosis, as investigated by flow cytometric analyses of Annexin V staining, was unaffected by contact with wild type PC cells or with IL30-overexpressing or IL30-deleted PC cells (Supplementary Fig. S3). Moreover, treatment of both HUVEC and HAEC with recombinant human (rh) IL30 significantly (Student's t-test: p < 0.0001) fostered their proliferation, as shown by MTT assay (Fig. 1C, D), and activation, as demonstrated by flow cytometric analyses of adhesion molecule (ICAM1, ICAM2, VCAM1 and P-selectin) expression, showing the upregulation of VCAM1 (CD106) and P-selectin (CD162) (Fig. 1E), which mediate leukocyte adhesion and can shape the TME [23]. Notably, endothelial cell proliferation assay of HUVEC, cocultured with IL30-overexpressing or IL30 gene knockout DU145 and PC3 cells, confirmed that tumor-derived IL30 stimulates EC proliferation, which was suppressed by coculture with IL30KO-PC cells, as shown by confocal microscopy and automated quantitative image analysis of CD31⁺ECs cocultured with EpCAM⁺PC cells (Fig. 1F, G, H). Finally, endothelial cell tube formation assay revealed that the number of capillaries generated from HUVEC and HAEC after 24 h of culture in Matrigel embedded with rhIL30 was significantly higher than the number of capillaries developed in control cultures (Student's t-test: p < 0.0001) (Fig. 1I, J, K, L), therefore demonstrating IL30's ability to promote vascular budding.

Since contact with PC cells stimulated endothelial proliferation and capillary bud formation, which were further improved by their overexpression of IL30, and impoverished by IL30 gene deletion in PC cells, we wondered whether PC—EC contact, and membrane-anchored IL30 of PC cells, had an impact on the endothelial expression of angiogenesis-related genes.

Angiogenesis PCR array (Supplementary Table S4) revealed that coculture with PC cells substantially affects the transcriptional profile of ECs (Fig. 2A, B). Following coculture with DU145, both HUVEC and HAEC overexpressed PROK2, PLG, CXCL9, TGFB2, FGF1, THBS2, TIMP3, CXCL10, EDN1, ANGPT2, JAG1, F3, ANG, EFNB2, MMP2, NOTCH4. Coculture with PC3 cells led, in both EC types, to the upregulation of TGFB2, CXCL9, ITGAV, CXCL10, IFNA1, ANG, TIMP3, IGF1 and EDN1, whereas only PTGS1 was downregulated.

The upregulation of ANG [24], CXCL10 [18], CXCL9 [25], EDN1 [26, 27], TGFB2 [28] and TIMP3 [29] was shared by both HUVEC and HAEC, after coculture with DU145 and PC3 cells (Fig. 2C). Therefore, contact with prostate cancer cells fostered endothelial expression of angiogenesis regulators, the majority with angiogenic effects, which was confirmed at protein level by western blot analyses (Fig. 2D, Supplementary Fig. S4).

To assess the role of tumor membrane-anchored IL30 in PC-endothelium crosstalk, we next investigated angiogenesis-related gene expression in ECs after coculture with IL30-overexpressing PC cells. Contact with IL30-DU145 strongly upregulated the expression of ITGAV, IGF1, TGFA, JAG1, CXCL1, CXCL10, HGF and EDN1, whereas it downregulated the expression of COL4A3 [30], in both HUVEC and HAEC (Fig. 3A). A wider range of genes, most of which endowed with angiogenic, angiocrine and inflammatory function, were upregulated in both HUVEC and HAEC following their coculture with IL30-PC3, such as TGFB2 [31], IGF1 [32], JAG1 [33, 34], CCL11/Eotaxin [35], NOS3 [36], FGF2 [37] and ENG/endoglin [38] (Fig. 3B).

The upregulation of CXCL10, EDN1, IGF1, along with that of ITGAV and JAG1 [39, 40], which was confirmed by flow cytometry, was shared by HUVEC and HAEC, after coculture with both IL30-DU145 and IL30-PC3 cells (Fig. 3C, D).

Intriguingly, analysis of the transcriptional profile of HUVEC, after contact with IL30 overexpressing DU145 or PC3 cells, revealed that, among the pro-angiogenic factors induced by this interaction, ANG [41] was the most upregulated (650 times in ECs cocultured with IL30-DU145, and 35 times in ECs cocultured with IL30-PC3 cells, whereas ITGAV upregulation ranked second, rising up to 594 times in ECs cocultured with IL30-DU145, and up to 24 times in ECs cocultured with IL30-PC3 cells) (Fig. 3E).

Since cytokine signaling pathways are commonly mediated through the phosphorylation of signal-dependent transcription factors, we carried out a phospho-kinase protein array using lysates of rhIL30-treated and untreated ECs (Fig. 3F, G). Seven proteins in HUVEC, including SRC, YES, GSK3α/β, and ten proteins in HAEC, including STAT3, STAT6, RSK 1/2, C-JUN, AKT, WNT and GSK3β, showed significant differences in phosphorylation after rhIL30 treatment. Most notably CREB [42, 43], GSK3β [44], HSP60 [45] and p53 [46], were significantly more phosphorylated in both HUVEC and HAEC (Fig. 3H, Supplementary Fig. S5), which can lead to endothelial dysfunction and regulate endothelial release of growth and inflammatory factors, as revealed by PCR array.

To assess whether IL30 directly affects the production of these factors, both EC types were treated with different concentrations of rhIL30. ELISA assay revealed that both HUVEC and HAEC constitutively released IGF1 (139.67 ± 2.31 and 283.67 ± 15.28 pg/ml, respectively), CXCL10 (4.46 ± 0.29 and 29.08 ± 1.84 pg/ml, respectively), EDN1 (88.20 ± 0.90 and 106.00 ± 1.30 pg/ml, respectively), and that treatment with rhIL30 (50–100 ng/ml) significantly increased their production and release (Fig. 4A, B). In addition, treatment of HUVEC with rhIL30 substantially increased the basal release of ANG (Fig. 4C), therefore confirming the ability of tumor membrane-anchored IL30 to regulate the endothelial production of angiogenesis factors. Flow cytometry showed that both HUVEC and HAEC expressed IGF1R, CXCR3 (all three isoforms A, B and alt were expressed, as shown in the Supplementary Fig. S6), EDNRA and EDNRB (Fig. 4D), and treatment with recombinant IGF1, CXCL10, and EDN1 significantly increased (ANOVA: p < 0.001) their proliferation (Fig. 4E, F), which was inhibited (ANOVA: p < 0.001) by the treatment with neutralizing anti-IGF1, anti-CXCL10, or anti-EDN1 Abs (Fig. 4G, H). The treatment of HUVEC with recombinant ANG, a potent inducer of new blood vessel formation that binds to high-affinity, yet to be identified, endothelial cell-surface receptors [41], increased their proliferation, which was suppressed by anti-ANG Abs (Fig. 4I, J).

Notably, IL30-induced endothelial hyperproliferation was substantially reduced (ANOVA: p < 0.0001) by neutralizing antibodies against each of the angiogenesis stimulating factors (IGF1, CXCL10, EDN1, ANG), which also inhibited spontaneous EC proliferation, due to their baseline production and release (ANOVA: p < 0.0001) (Fig. 4K-Q). IGF1, CXCL10, EDN1 and ANG cooperate substantially in mediating IL30-driven endothelial hyperproliferation and are most likely involved in IL30 driven in vivo angiogenesis, since IL30 overexpressing tumors developed after xenograft implantation of IL30-PC3 or IL30-DU145 cells, which have demonstrated overgrowth and metastatic behavior [15], showed a prosperous vascularity, hyperproliferation and strong expression of these angiogenesis regulators when compared to control tumors, as reported in ref 15 and confirmed by data on PC3 tumor xenografts (Supplementary Table S5, Fig. 4R, S, T).

To answer the question of whether abrogation of IL30 expression in PC cells could affect the transcriptional profile of contiguous ECs, we performed angiogenesis PCR array in ECs cocultured with IL30KO-DU145 or IL30KO-PC3 cells.

Compared to coculture with wild type DU145 cells, coculture with IL30KO-DU145 cells drastically inhibited, in both HUVEC and HAEC, the expression of a wide range of proangiogenic genes (Fig. 5A), including IGF1, EDN1, CXCL10 and ITGAV, which were found to be upregulated in ECs cocultured with IL30 overexpressing PC cells (Fig. 3A, B, C).

Contact with IL30KO-PC3 cells led to the suppression of different proangiogenic genes, including VEGFA, TGFA, ANGPT2 and ANGPTL4 in both HUVEC and HAEC (Fig. 5B).

Only the inhibition of CXCL6, IGF1, TGFA and THBS2 expression was shared by HUVEC and HAEC following their culture with IL30KO-DU145 or IL30KO-PC3 cells and confirmed at the protein level by WB or immunopathological examination of IL30 deficient tumors induced by IL30 gene deleted PC cell implantation in NSG mice (Supplementary Table S6; Fig. 5D, E, F, G, H and Supplementary Fig. S7). These findings suggest that the effects, on the endothelial gene signature, of the interaction with IL30 knockout PC cells is tumor specific.

The finding that contacts with PC cells and, even more so, their contact with IL30 overexpressing or deficient PC cells, could regulate the angiogenic profile of ECs, led us to wonder whether the endothelial-cancer cell crosstalk could also have impact on the immune profile and oncogenic potential of PC cells. To answer this question, the expression of immunoregulatory and PC driver genes was investigated by PCR array (Supplementary Tables S7 and S8) in wild type and IL30-overexpressing DU145 and PC3 cells cocultured with normal ECs.

Coculture with HUVEC determined, in both DU145 and PC3 cells, an impressive upregulation of proinflammatory and immunoregulatory genes, especially BCL2, CCL21, CCL22, CCR1, CSF3, FASL, IL1B, IL4 and NOS2 (Fig. 6A, C) and a consistent upregulation of prostate cancer driver genes, that included DAXX, FASN, HMGCR, IL6, MKI67, PDPK1, PES1, SOX4 and SREFB1 (Fig. 6B). Among prostate cancer driver genes, the tumor suppressor gene ZNF185 was downregulated, whereas GNRH1, PPP2R1B and TP53 were upregulated in both DU145 and PC3 cocultured with HUVEC versus cancer cells cultured alone. Furthermore, when overexpressing IL30, DU145 and PC3 cells showed, after coculture with ECs, a higher and broader expression of both inflammation and PC driver genes compared to monocultured IL30-DU145 or IL30-PC3 cells (Fig. 6D, E, F). The crosstalk with ECs, that had been stimulated by IL30 overexpression in PC cells, also shaped the transcriptional profile of PC cells by expanding the range and level of expression of immunity genes, such as those coding for chemokines, as CCL28, CCL4 and CCL5; chemokine receptors, as CCR2, CCR7, CCR9 and CXCR4; cytokines and growth factors, as IL10, IL13, IL17A, EGF and VEGFA (Fig. 6D). Similarly, after their culture with ECs, IL30-overexpressing PCs showed a dramatic upregulation of a wide range of oncogenes, which included CCND2, EGR3, IGFBP5, KLK3, PDLIM4, PTGS1 and SHBG and a few tumor suppressors, such as GPX3, FOXO1, MAX and NKX3-1 (Fig. 6E, F).

Since in clinical PC samples cancer cells are in close contact with the endothelium, to assess whether the experimental findings have clinical relevance, we investigated in patient-derived PC samples the expression levels of IL30 along with that of immunoregulatory and PC driver genes, which were regulated in both IL30 overexpressing DU145, and PC3 cells after coculture with ECs when compared to wild type DU145, and PC3 cells, respectively, cocultured with ECs (Fig. 7A, B). Bioinformatic analysis was performed by using the “Prostate Cancer Transcriptome Atlas” (PCTA), the largest of the publicly available databases, which contains RNA-Seq data obtained from 1116 clinical PC specimens, with annotation of Gleason score, collected from 38 PC patient cohorts [47].

Among the PC driver genes that were found upregulated in IL30 overexpressing PC cells, after coculture with ECs (Fig. 7A), genes coding for Galectin 4 (LGALS4), Gonadotropin Releasing Hormone 1 (GNRH1), and Sex Hormone Binding Globulin (SHBG) correlated with IL30 expression also in clinical PC samples, as assessed by bioinformatics of RNA-Seq data from the PCTA collection. Specifically, expression of LGALS4 and GNRH1 (ρ = 0.36 and ρ = 0.33, respectively; p < 0.01), and especially that of SHBG (ρ = 0.47; p < 0.01) positively correlated with IL30 expression.

Among the inflammation and immunity genes that were found to be upregulated in IL30 overexpressing PC cells, after coculture with ECs (Fig. 7B), genes coding for Nitric Oxide Synthase 2 (NOS2), Tumor necrosis Factor alpha (TNFA), C-X-C chemokine receptor type 5 (CXCR5) and Interleukin-12B (IL12B) correlated with IL30 expression also in clinical PC samples from the PCTA collection. Specifically, expression of NOS2, TNFA, CXCR5 (ρ = 0.38, ρ = 0.37 and ρ = 0.36, respectively; p < 0.01), and primarily that of IL12B (ρ = 0.47; p < 0.01) positively correlated with IL30 expression. Moreover, immunohistochemical analysis confirmed, in vivo, overexpression of the aforementioned inflammation and cancer driver genes (IL12B, SHBG, TNFA, LGALS4) in IL30-overexpressing DU145 and PC3 tumors, compared to control tumors (Fig. 7C, Supplementary Fig. S8 and Supplementary Table S9).

Since IL12B (p40 subunit of the Interleukin 12 family of cytokines) has been found to protect PC cells from apoptosis [48], we next investigated the relationship between the expression of IL12B and that of apoptotic-related genes (Supplementary Table S2) in PC samples. The Spearman rank sum of RNA-Seq data from the PCTA database, demonstrated an inverse correlation between the transcriptional expression of IL12B and that of the genes involved in the apoptotic signaling pathway (ρ = -0.40; p < 0.01). Expression of apoptosis-related genes also showed an inverse correlation with the expression of SHBG (ρ = -0.33; p < 0.01) in PC sample database. The inverse correlation was stronger (ρ = -0.44; p < 0.01) when the combined expression levels of IL12B and SHBG were considered, suggesting that inhibition of apoptosis may be a further downstream mechanism that mediates the tumor promoting activity of IL30.

To assess at the protein level the correlations between gene expressions resulting from both experimental and bioinformatic findings, we next performed immunopathological analyses of tumor samples obtained from 80 patients (a cohort size which allowed the detection of a statistically significant correlation between two genes, with an 85% power and a 5% significance level) who underwent surgery for PC and were selected from our institutional Biobank by matching for Gleason score with PC patients of the PCTA collection (Table 1). Morphometric analysis confirmed, at protein level, the positive correlation between expression of IL30 and that of IL12B and SHBG (ρ = 0.50 and ρ = 0.52, respectively; p < 0.01) and, to a lesser extent, between IL30 positivity and immunostainings for LGALS4 (ρ = 0.32), GNRH1 (ρ = 0.31), NOS2 (ρ = 0.36), TNFA (ρ = 0.33) and CXCR5 (ρ = 0.33) (p < 0.01), therefore substantiating the biostatistical findings (Fig. 7D, Supplementary Fig. S9, the quantification of the immunohistochemical expression of inflammation and immunity genes, and prostate cancer driver genes, for each patient, is reported in the Supplementary Table S10) and highlighting the intricate network of immunoregulatory and cancer driver genes correlated with IL30 expression in the PC microenvironment.