Suppression of endothelial miR-22-3p mediates non-small cell lung cancer cell-induced angiogenesis

MicroRNAs (miRNAs) expressed in endothelial cells (ECs) are powerful regulators of angiogenesis, which is essential for tumor growth and metastasis. Here, we demonstrated that miR-22-3p (miR-22) is preferentially and highly expressed in ECs, while its endothelial level is significantly down-regulated in human non-small cell lung cancer (NSCLC) tissues when compared to matched non-tumor lung tissues. This reduction of endothelial miR-22 is induced by NSCLC cell-secreted tumor necrosis factor (TNF)-α and interleukin (IL)-1β. Endothelial miR-22 functions as a potent angiogenesis inhibitor that inhibits all the key angiogenic activities of ECs and consequently NSCLC growth through directly targeting sirtuin (SIRT) 1 and fibroblast growth factor receptor (FGFR) 1 in ECs, leading to inactivation of AKT/mammalian target of rapamycin (mTOR) signaling. These novel findings provide insight into the molecular mechanisms of NSCLC angiogenesis and indicate that endothelial miR-22 represents a potential target for the future anti-angiogenic treatment of NSCLC.


Introduction
Angiogenesis, i.e. the formation of new blood vessels from pre-existing ones, is essential for 49 tumor growth and metastasis. Accordingly, excessive angiogenesis is a poor prognostic 50 indicator for the aggressiveness of different cancer types, such as non-small cell lung cancer 51 (NSCLC) (1). Tumor angiogenesis is tightly regulated by the balance between pro-and anti-52 angiogenic factors, which involves the dynamic communication between tumor cells and 53 endothelial cells (ECs). Tumor cells are capable of releasing different pro-angiogenic factors, 54 such as vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF, 55 FGF2), epidermal growth factor (EGF), tumor necrosis factor (TNF)-α, interleukin (IL)-1β, 56 IL-6 and 3). The binding of these factors to their receptors located on ECs activates 57 pivotal downstream angiogenesis-related signaling pathways, such as phosphoinositide 3 58 kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) signaling (4). Consequently, 59 ECs are stimulated to degrade their basement membrane, proliferate, migrate toward tumor 60 cells and interconnect with each other to form new microvascular networks (2, 4). 61 Previous studies have shown that sirtuin (SIRT) 1 plays a crucial role in the regulation 62 of angiogenesis (5). SIRT1 is a prototype member of the sirtuin family of nicotinamide 63 adenine dinucleotide-dependent class III histone deacetylases. Loss of SIRT1 results in a 64 significant reduction of EC sprouting and branching activity (5). Moreover, endothelial SIRT1 65 deletion impairs angiogenesis within ischemic hindlimbs and the kidney (5, 6). The pro-66 angiogenic effect of SIRT1 is most probably mediated by some of its substrates. In fact, it has 67 been reported that SIRT1 deacetylates AKT, which binds to phosphatidylinositol (3,4,5)-68 triphosphate, leading to the activation of the AKT/mTOR pathway (7). In addition, SIRT1 69 deacetylates the forkhead transcription factor FOXO1 and, thus, suppresses its anti-  (10). which 81 is located on chromosome 17p13 and highly conserved among metazoans (11), has been 82 reported to be also expressed in different types of ECs (12). However, its role in regulating 83 tumor angiogenesis remains elusive. 84 In the present study, we analyzed the regulation of endothelial miR-22 by NSCLC cells. 85 We then systematically investigated the function of miR-22 in basic angiogenic processes, 86 including EC proliferation, migration and tube formation. The observed anti-angiogenic 87 action of miR-22 was further confirmed in an ex vivo mouse aortic ring assay and an in vivo 88 Matrigel plug assay. In addition, we studied the effects of endothelial miR-22 on tumor 89 angiogenesis and growth in a mouse flank tumor model. Finally, mechanistic analyses 90 identified SIRT1 and FGFR1 as functional targets of miR-22 in ECs.

93
Endothelial miR-22 is down-regulated in human NSCLC tissues 94 In a first step, ECs lining the blood vessels in tumor tissues and matched adjacent non-tumor 95 lung tissues from 12 patients with lung adenocarcinoma were retrieved by means of laser 96 capture microdissection (LCM). By a small-scale screening using real-time PCR we identified 97 miR-22 to be significantly down-regulated in ECs isolated from tumor tissues when compared 98 to those isolated from matched non-tumor lung tissues ( Figure 1A). 99 Of note, miR-22 was found to be preferentially and highly expressed in both types of 100

108
Since tumor cells are capable of stimulating the angiogenic activity of ECs by both direct cell-109 cell contact and paracrine signaling, we next utilized a contact co-culture system to 110 investigate how the expression of miR-22 in HDMECs is regulated by NSCLC cells. After 24 111 h of either culturing HDMECs alone or co-culturing them with NCI-H460 or NCI-H23 cells, 112 HDMECs were isolated using CD31 magnetic beads. The purity of isolated HDMECs was 113 approximately 99% and 90% in the HDMECs mono-culture and co-culture group, 114 respectively, as assessed by flow cytometry. Real-time PCR assays revealed a 25% and a 18% 115 reduction of miR-22 expression in HDMECs co-cultured with NCI-H460 cells and NCI-H23 116 cells, when compared to HDMEC mono-culture ( Figure 1C). In an additional set of 117 experiments, we co-cultured HDMECs with NSCLC cells, however, without contact between 118 these two cell types in a transwell plate. Interestingly, this non-contact co-culture with NCI-119 H460 cells caused a 35% decrease in the miR-22 expression level of HDMECs ( Figure 1D), 120 indicating that soluble factors secreted by the tumor cells contribute to the down-regulation of 121 endothelial miR-22. This finding was confirmed by the co-culture of HDMECs with NCI-H23 122 cells, which also significantly reduced the endothelial expression of miR-22 by 31% ( Figure   123 1D). 124 In order to identify individual factors mediating the NSCLC cell-induced reduction of 125 endothelial miR-22, HDMECs were stimulated with the growth factors VEGF, bFGF and 126 EGF as well as the pro-inflammatory cytokines TNF-α, IL-1β and IL-6. Real-time PCR 127 analyses revealed that the expression of miR-22 is significantly suppressed by TNF-α and IL-128 1β, but not affected by VEGF, bFGF, EGF and IL-6 stimulation ( Figure 1E). Given the fact 129 that both TNF-α and IL-1β are upstream inducers of nuclear factor (NF)-κB, which promotes 130 or represses the transcription of a broad spectrum of genes and miRNAs (13, 14), we then 131 investigated whether NF-κB inhibits the transcription of miR-22 in ECs. For this purpose, 132 HDMECs were exposed to the NF-κB inhibitor Bay 11-7082 (Bay) for 24 h. This resulted in a 133 2-fold increase of miR-22 expression when compared to vehicle-treated controls ( Figure 1F), 134 indicating that this miRNA is transcriptionally repressed by NF-κB. 135 To investigate whether NF-κB mediates the down-regulation of endothelial miR-22 136 induced by NSCLC cells, we assessed the activation status of NF-κB in HDMECs cultured 137 alone or co-cultured with NCI-H460 cells without contact. By means of immunofluorescence, 138 we demonstrated that the nuclear translocation of p65, a main subunit of NF-κB, is 139 significantly enhanced in HDMECs co-cultured with tumor cells ( Figure 1G and H). 140 Importantly, blockade of NF-κB signaling with Bay completely reversed the reduction of 141 endothelial miR-22 induced by non-contact co-culture with NCI-H460 cells ( Figure 1I). . 168 In addition, we performed a tube formation assay to investigate the function of miR-22 169 in regulating the tube forming activity of HDMECs. Transfection with miR-22m markedly 170 reduced the number of newly developed tube meshes by 76% when compared to NCm-transfected controls ( Figure 2G and H). In contrast, miR-22i significantly augmented EC tube 172 formation by 64% ( Figure 2I and J).

174
Endothelial miR-22 suppresses angiogenesis ex vivo and in vivo 175 To elucidate whether miR-22 is involved in endothelial sprouting, we performed an ex vivo 176 mouse aortic ring assay. We found that the area of vascular sprouting from aortic rings is 177 significantly decreased by transfection with miR-22m ( Figure 3A and B) and significantly 178 increased by transfection with miR-22i ( Figure 3C and D). 179 To confirm our in vitro findings, we performed an in vivo Matrigel plug assay. Matrigel 180 plugs containing miR-22m-transfected HDMECs exhibited a 58% reduction of the 181 microvessel density 7 days after implantation when compared to those containing NCm-182 transfected controls ( Figure 3E and F). In contrast, plugs containing miR-22i-transfected cells 183 presented with a 42% higher microvessel density than plugs containing NCi-transfected cells 184 ( Figure 3G and H).

207
To identify the functional targets of miR-22 that mediate its anti-angiogenic effects in ECs, 208 we first analyzed the predicted human target genes of miR-22 according to the algorithms of 209 miRDB and TargetScan. We detected 5 genes that are involved in angiogenesis and have not 210 been validated as miR-22 targets, which encode tumor necrosis factor receptor (TNFR) 2, 211 vascular endothelial zinc finger (VEZF) 1, transforming growth factor beta-activated kinase 212 (TAK) 1, serine-arginine protein kinase (SRPK) 1 and protein kinase C beta (PRKCB). 213 However, none of these genes was down-regulated in miR-22m-transfected HDMECs when 214 compared to NCm-transfected controls ( Figure 5A). means of a WST-1 assay, we found that 10-50 µM EX and 50-500 nM PD significantly 245 reduce the viability of HDMECs after 3 days of treatment ( Figure 6A and B). Accordingly, to 246 avoid cytotoxic effects of these compounds, we chose a minimal effective dose of each 247 inhibitor, i.e. 10 µM EX and 50 nM PD, for the following WST-1, scratch wound healing and 248 tube formation assays. These functional analyses revealed that exposure to EX and PD 249 completely reverses miR-22i-promoted HDMEC viability, migration and tube formation 250 ( Figure 6C-E). 251 Furthermore, we analyzed whether miR-22 functions through suppressing AKT/mTOR 252 signaling, which is a common down-stream pathway of SIRT1 and FGFR1, using the highly 253 specific AKT inhibitor MK-2206 (MK). In a previous publication (17), we found that 5-40 254 µM MK-2206 significantly reduces HDMEC viability after 3 days of incubation. 255 Accordingly, miR-22i-transfected HDMECs were exposed to 5 µM MK-2206 followed by 256 WST-1, scratch wound healing and tube formation assays. By this, we could demonstrate that 257 inhibition of AKT completely counteracts miR-22i-enhanced HDMEC viability, migration   The transwell migration assay was performed as previously described (32). Briefly, Statistical comparisons between two groups were made by the paired Student's t-test (for the 625 analysis of patient samples) or the unpaired Student's t-test using GraphPad Prism 9. 626 Statistical comparisons between multiple groups were made by one-way ANOVA followed 627 by the Tukey's multiple comparisons test using GraphPad Prism 9. All data were expressed as 628 means ± SEM. A value of P<0.05 was considered significant.