The metal-nonoate Ni(SalPipNONO) inhibits in vitro tumor growth, invasiveness and angiogenesis

Nitric oxide (NO) exerts conflicting effect on tumor growth and progression, depending on its concentration. We aimed to characterize the anti-cancer activity of a new NO donor, Ni(SalPipNONO) belonging to the class of metal-nonoates, in epithelial derived tumor cells, finally exploring its antiangiogenic properties. Tumor epithelial cells were screened to evaluate the cytotoxic effect of Ni(SalPipNONO), which was able to inhibit cell proliferation in a dose dependent manner, being more effective than the commercial DETA/NO. The human lung carcinoma cells A549 were chosen as model to study the anti-cancer mechanisms exerted by the compound. In these cells, Ni(SalPipNONO) inhibited clonogenicity and cell invasion, while promoting apoptosis. The antitumor activity was partly due to NO-cGMP dependent pathway, contributing to reduced cell number and apoptosis, and partly to the salicylaldehyde moiety and reactive oxygen species (ROS) activated ERK1/2 signaling converging on p53 dependent caspase-3 cleavage. An additional contribution by downstream cycloxygenase-2 (COX-2) derived cyclopentenones may explain the tumor inhibitory activities. As NO has been described to affect tumor angiogenesis, we checked this activity both on tumor and endothelial cell co-cultures and in Matrigel in vivo assay. Our data document that Ni(SalPipNONO) was able to both reduce angiogenic factor expression by tumor cells acting on hypoxia inducible factor-1α (HIF-1 α) level, and endothelial cell functions related to angiogenesis. Collectively, these data confirm the potential use of NO donors and in particular Ni(SalPipNONO) acting through multiple mechanisms, as an agent to be further developed to be used alone or in combination with conventional therapy.


INTRODUCTION
From the first studies on the effect of nitric oxide (NO) in cancer biology, this mediator emerged as a biphasic modulator, behaving both as an antineoplastic and proneoplastic stimulus [1]. The bimodal actions of NO can be explained through the duration of NO exposure, the cellular microenvironment, NO flux, tumor cell proliferation rate, occurrence of oxidizing and reducing processes [1].
The established biochemical/cellular events elicited by NO against tumor development are essentially inhibition of cell proliferation and proapoptotic events, and vascular effects including anti-angiogenesis. The

Research Paper
proapoptotic events described in the literature relate to p53 upregulation and accumulation, degradation of antiapoptotic mediators, mitochondrial membrane permeability, and induction of cytochrome c release [2][3][4]. In cancer cells, NO released by NO donors or nitric oxide synthase (NOS) has been suggested to activate p53 via DNA damage by peroxynitrite (ONOO -) [5][6][7].
The use of NO donors can improve vascular flow, and anticancer drug delivery in hypoxic tissue, favoring the penetration of chemotherapy in tumor tissue and improving their cytotoxic effects [8][9][10]. Indeed, an increase in response to radiotherapy [10,11] and chemotherapy [12,13] has been reported.
Tumor tissue is characterized by low oxygen tension, a condition that promotes the activation and stabilization of hypoxia inducible factor-1α (HIF-1α) which, in turn, controls the transcription of vascular endothelial growth factor (VEGF), thus promoting angiogenesis, tumor growth and metastasis [14,15]. NO has been reported to inhibit the expression of HIF-1α through the activation of HIF-1-prolyl hydroxylases and its proteasomal degradation [16][17][18][19]. NO, by reducing HIF-1α dependent VEGF levels, at the end can improve the delivery of antitumor drugs, through vascular normalization and reversion of the oncotic pressure gradient [20].
Recently, a new family of metal-nonoates has been developed [21] and characterized for their potential use in cardiovascular diseases, characterized by endothelial dysfunction, obtaining a vascular protective effect at nanomolar concentrations [22,23]. Here, we have evaluated the antitumor activity of a member of this class, Ni(SalPipNONO), assessing the antitumor efficacy in two epithelial derived tumor cells, A549 and HT29, representative of lung and colon carcinoma, respectively. Ni(SalPipNONO) was characterized for different mechanisms related to tumor hallmarks as well as for its antiangiogenic effects on tumor and endothelial cells.

Antitumor effects and mechanisms of action of Ni(SalPipNONO)
To test the effect of novel NO donor, human lung carcinoma cells A549 cells were exposed for 72 h to Ni(SalPipNONO) and DETA/NO used in a wide range of concentrations (0.001-1 mM) and cell viability was assessed by the MTT assay. The experiment was performed in 0.1 and 2% FBS ( Figure 1A and 1B). Ni(SalPipNONO), compared with equimolar concentrations of DETA/ NO, was more effective in reducing cell number, in particular in the range 0.1-1 mM. The EC 50 for Ni(SalPipNONO) were 0.26 and 0.37 mM in 0.1 and 2% serum, respectively. To assess the antiproliferative effect of the nonoate, BrdU incorporation assay were performed after 24 h of Ni(SalPipNONO) treatment in 0.1 and 2% FBS ( Figure 2C). In both experimental conditions, the viability of A549 cells was less than 50% after exposure to 1 mM of NO donors. These experiments show that Ni(SalPipNONO) exerted its antiproliferative effects at doses near 0.5 mM, while at 1 mM it revealed a cytotoxic action.
Similar results were obtained with the human colon adenocarconma cells HT29 (Supplementary Figure 1A and 1B).
When tested on normal cells, namely HaCaT keratinocytes, Ni(SalPipNONO) exerted antiproliferative action only on sparse cells exposed to 2% serum (Supplementary Figure 2A). Interestingly, when the nonoate was tested on confluent cells exposed to low serum concentration, a more physiological condition, it did not inhibit cell proliferation (Supplementary Figure  2B) until 1 mM of drug.
Next, the cellular and biochemical characterization of Ni(SalPipNONO) antitumor activity was performed in A549 cells, using 0.5 mM concentration of the NO donor and control molecules. This concentration corresponded to a significant antiproliferative effects (on proliferating tumor cells) which allowed to study biochemical effects.
Beside cytotoxic/antiproliferative activity, we have studied the capability of metal-nonoate to interfere with clonogenicity of A549 cells. Data showed that at 0.5 and 1 mM of Ni(SalPipNONO) the surviving fraction was reduced of >90% with respect to basal, while DETA/NO exerted a modest, non-significant, effect (Figure 2A and 2B).
It is well known that metastasis of cancer cells involves cell invasiveness [24]. By using the Boyden chamber and gelatin coated filters, we have investigated the activity of NO donors on cell migration. As shown in Figure 2C, the test has been carried out in the presence of low and high concentration of an unspecific chemoattractant agent, i.e. serum. In both conditions, after 18 h of exposure to 0.5 mM Ni(SalPipNONO), we could detect halving of cells migrated across filter toward the lower chamber. The inhibitory effect of DETA/NO was evident only when cells were stimulated by a serum gradient.
From all these data, the metal-nonoate Ni(SalPipNONO) at sub-millimolar concentration showed an antiproliferative effect on lung cancer cells, accompanied by reduced clonogenicity and invasiveness.
We then evaluated the mechanisms responsible for the tumor inhibitory effects.
First, we evaluated the involvement of the classical soluble guanylate cyclase (sGC)/cGMP pathway by the use of ODQ. The preincubation with ODQ (10 µM, 30 min before Ni(SalPipNONO) only partially reverted the NO donor cytotoxic effect ( Figure 3A), suggesting that other pathways or chemical components of the molecule could be responsible of the antitumor actions. Oxidative stress and ROS production have been described to contribute to the cytotoxic effect of high doses on NO, due to its radical nature [6,7]. We measured ROS levels in A549 cells stimulated with the metal-nonoate in different serum concentrations by means of the fluorophore DCFH2-DA and we found a significant burst in ROS production that was abolished by NAC pretreatment ( Figure 3B).
Next, to evaluate the portion of molecule responsible of cytotoxic activity, we have tested different compounds with or without the NONO or the salicylaldehyde moiety at equimolar concentration. The compounds with salicylaldehyde group exerted the greater cytotoxic action (probably due to its metabolism in salicylic acid), with a synergetic effect with NONO group ( Figure 3C). Ultimately, we can speculate that both the NONO group and salicylaldehyde moiety concur to the cytotoxic action. Since the compound devoid of NONO group did not create an oxidative environment in cells ( Figure 3B), a double mechanism of cytotoxic action could be hypothesized: the NONO group induces oxidative stress, and salicylaldehyde exerts an additional mechanism.
Since COX-2 derived prostanoids can influence tumor development, and inflammation is one of the tumor hallmarks [24], we have evaluated the activity of Ni(SalPipNONO) on COX-2 protein levels. As reported in Figure 4A, COX-2 was upregulated by 0.5 mM concentration of the metal-nonoate, but not by DETA/NO. From a mechanistic point of view, the inhibition of COX-2 activity by NS398 partially reverted the metal-nonoate cytotoxic effect ( Figure 4B), hypothesising that inhibitory prostanoids could contribute to antitumor activity. Indeed, cyclopentenones as 15d-PGJ 2 have been reported to inhibit cell growth and induce apoptosis in various tumors [25][26][27][28][29]. In our tumor model, when cells were exposed to exogenous 15d-PGJ 2 , a dose dependent inhibition of cell chemotaxis toward 10% serum was detected ( Figure  4C), suggesting a contribution of endogenously processed inhibitory prostanoids.
The capability of NO to induce apoptosis is widely reported in literature [1,2,5,6], thus to understand the mechanisms of cell number reduction promoted by Ni(SalPipNONO), the expression of apoptotic markers was evaluated. Exposure of cells to Ni(SalPipNONO) (0.5 mM, 15 min) doubled cytochrome c levels ( Figure  5A), maybe through the reactive species generated by NO (like peroxynitrite and ROS) that cause opening of the permeability transition pore of mitochondria [2,30]. Moreover, it is known that ROS activate ERK1/2 [31] and   in A549 cells Ni(SalPipNONO), but not DETA/NO, strongly increased pERK1/2 after 30 min of incubation ( Figure 5B). ERK1/2 on its turn upregulates p53 [32][33][34] and in our experiments p53 maximum level was obtained after 60 min of incubation with the metal-nonoate ( Figure 5C). Indeed, the preincubation of the cells with MEK inhibitor U0126 interfered with p53 upregulation ( Figure 5D).
The ultimate event downstream to cytochrome c release is caspase 3 activation that leads to DNA cleavage and apoptosis [35]. Our data show a maximum production (triple respect basal) of cleaved-caspase 3 after 4 h of exposure with Ni(SalPipNONO) ( Figure 5E), an effect which was completely abolished by the preincubation with the p53 inhibitor PT-α ( Figure 5F).
From all these data it results that Ni(SalPipNONO) is able to impair cell migration and reduce cell number and survival by activating the apoptosis pathway that passes through ROS production and COX-2 cyclopentenone activation, beside the partial involvement of the sGC/ cGMP conventional signaling.

Antiangiogenic activity by Ni(SalPipNONO) on tumor and endothelial cells
HIF-1α expression and proangiogenic factor levels are correlated with an increased risk of mortality in several types of carcinoma [14]. In A549 cells, we could find HIF-1α detectable levels already in normoxic condition ( Figure 6A). When A549 cells were treated for 24 h with Ni(SalPipNONO), there was a substantial decrease of HIF-1α expression and a consequent reduction of VEGF levels ( Figure 6A). These results demonstrate that the metalnonoate reduces the angiogenetic signals in tumor cells.
We then assessed whether the antiangiogenic property was directly induced in endothelial cells. HUVEC were exposed to VEGF (20 ng/ml) in the presence of Ni(SalPipNONO). Interestingly, at 0.1 mM the NO donor completely abolished VEGF induced proliferation ( Figure  6B). The antiangiogenic activity of Ni(SalPipNONO) was evident also in an in vivo angiogenesis assay as the subcutaneous Matrigel plug implant. After 10 days, the presence of metal-nonoate abolished the vascularization produced by VEGF, as documented by representative pictures and hemoglobin content in the plugs ( Figure 6C and 6D).
Co-culture experiments with tumor and endothelial cells were set up to strengthen the above results. Endothelial cell organization on Matrigel layers was evaluated in the presence of tumor cells grown on transwell inserts and treated or not with Ni(SalPipNONO).
In the presence of untreated A549, in 18 h HUVEC formed net-like structures, which were impaired when tumor cells were treated with the NO donor ( Figure 7A).Within the same time of incubation, a reduction in VEGF expression  Figure 7B).
These data in the whole demonstrate that the antiangiogenic activity of Ni(SalPipNONO) is both direct on endothelial cells and indirect on tumor cells which were not able to upregulate VEGF, probably through a COX-2/ cyclopentenone dependent mechanism.

DISCUSSION
Here we describe for the first time the antitumor efficacy of novel metal-nonoates, in particular Ni(SalPipNONO) on A549 tumor cells, characterizing the cellular and biochemical profile of the NO donor. We have found the cytotoxic activity of Ni(SalPipNONO) at doses higher than 0.1 mM, accompanied by inhibition of tumor clonogenicity and invasiveness, hallmarks typical of malignant tumors. From a mechanistic point of view, multiple pathways are responsible for the antitumor and proapoptotic events: sGC/cGMP activation; ROS production/p-ERK1/2 and cytochrome c/p53 pathway; COX-2/cyclopentenone contribution.
The influence of the conventional sGC/cGMP pathway is only marginal, since the block of the cascade with ODQ only partially reverted the cytotoxic effect of the metal-nonoate.
Here we demonstrate that Ni(SalPipNONO) promotes ERK1/2 phosphorylation, possibly through NO associated-ROS production. NO-induced oxidative stress influences MAPK dependent p53 upregulation and activation ultimately leading to caspase-3 cleavage. This effect is typical of Ni(SalPipNONO), since DETA/ NO has lower capability to activate the MAPK pathway and apoptosis signalling and the compound devoid of the NONO group does not elicit an oxidative burst.
It is known that ROS mediated activation of ERK1/2 is able to activate the apoptotic pathway by antitumor agents [31]. ERK is part of the MAPK superfamily, and is well known for its ability to control cell survival in response to external stimuli [33,44]. Several reports have found more complex roles for ERK pathway, in which the increase of ERK activity might promote apoptosis in Pictures are representative plugs out of 3, while the graph (D) reports hemoglobin content quantified by Drabkin reagent (n = 3 plugs). *** p < 0.001 vs VEGF alone. specific environments. Protein kinase pathways such as the MAPK pathway are major oxidative stress-sensitive pathways in most cell types [45]. In particular, ERK is selectively activated in neuronal and renal epithelial cells upon exposure to oxidative stress and toxic agents such as cisplatin, and inhibition of the ERK pathway has been reported to block apoptosis [46]. Several studies have reported that curcumin potentiates ROS-dependent ERK activation and lethality in irradiated human cervical tumor cells [47] and that cisplatin-induced ERK activation is partly mediated through ROS generation [48]. In the current study, nonoate treatment induces a ROS burst and significantly increases ERK1/2 phosphorylation in A549 cancer cells. These results seem to be consistent with several earlier studies, in which increased ERK activity by reactive nitrogen species was linked to the induction of cell death [49].
The inhibitor of p53 PT-α blocked the apoptosome-mediated processing and activation of caspase-9 and -3 without interfering with the activation of mitochondria [50]. Our data show that apoptosis occurs via a p53-dependent mechanism that takes place upstream of mitochondria and involves cytochrome c release. Indeed, it is known that the reactive species generated by high levels of NO (both peroxynitrite and ROS) cause opening of the permeability transition pore of mitochondria [2,30].
We have then evaluated the contribution of endogenous prostanoid pathway in the anti-tumor and antiangiogenic effect of the metal-nonoate.Among the COX-2 derived prostanoids, the pro-inflammatory PGE 2 has a predominant role in promoting tumor growth [51]. However, both LOX-and COX-derived products can act as endogenous ligands of anti-proliferative and antitumorigenic receptor(s), as PPAR-γ. The cyclopentenone prostaglandins PGA 2 , PGA 1 , and PGJ 2 are formed by dehydration within the cyclopentane ring of PGE 2 , PGE 1 , and PGD 2 . PGJ 2 is metabolized further to yield D12-PGJ 2 and 15-deoxy-D12,14-PGJ 2 (15d-PGJ 2 ). Various compounds within the cyclopentenone prostaglandin family possess potent anti-inflammatory, anti-neoplastic, and anti-viral activity. Among cyclopentenones, 15d-PGJ 2 has attracted our attention.15d-PGJ 2 , given exogenously, reproduces the antitumor activity of the metal-nonoate. 15d-PGJ 2 has been show to inhibit angiogenesis via suppression of pro-inflammatory enzymes and cytokines, even if stimulatory functions on tumor angiogenesis have been reported [52]. In human umbilical vein endothelial cells 15d-PGJ 2 induced apoptosis via PPAR γ-dependent mechanisms [53]. The cyclopentenone involvement leads us to investigate the mechanisms related to the antiangiogenic activity of the metal-nonoate. Our in vitro and in vivo results document the antiangiogenic activity of Ni(SalPipNONO). Interestingly, the antiangiogenic activity of Ni(SalPipNONO) is both direct on endothelial cells and indirect on tumor cells via upregulation of VEGF. In the first ones it directly abrogates VEGF induced proliferative effect, while in tumor cells the nonoate inhibits HIF-1α dependent VEGF upregulation. An additional contribution of cyclopentenones as 15d-PGJ 2 in preventing VEGF expression is here demonstrated.
A direct antiangiogenic activity by exogenous NO has been reported with other NO donors. In vitro assay with cultured endothelial cells revealed that the NO hybrid compound NCX-4016, significantly inhibited angiogenesis in a dose-dependent manner, with almost complete inhibition at a 100 μM concentration [54]. And recently, alteration of the expression of angiogenic genes has been described in hepatoma cells in response to NO donors, in particular an increase in thrombospondin-1 and tissue inhibitors of metalloprotease-1 [55], known inhibitors of angiogenesis and tumor cell migration.
In conclusion, these data confirm the potential use of NO donors and in particular Ni(SalPipNONO) acting through multiple mechanisms, as chemotherapeutic agents to be further developed in order to be used alone or in combination with anticancer conventional therapy. The results of a randomized phase II clinical trial have been reported, documenting the feasibility of the concurrent use of NTG with chemotherapy and radiotherapy in locally advanced non-small cell lung cancer to increase chemoand radiosensitivity with an acceptable toxicity profile [56]. Interestingly, the overall survival was associated to reduced levels of circulating VEGF. Accordingly, a retrospective study suggested that application of NTG plus docetaxel and carboplatin in patients with operable lung adenocarcinoma increases the response with decreased expression of HIF-1α and VEGF [57], supporting an antiangiogenic activity.

BrdU incorporation assay
Cell proliferation was determined by 5-bromo-2′-deoxy-uridine (BrdU) incorporation using a chemioluminescence ELISA according to the manufacturer's instructions (#11669915001 Roche Diagnostic S.p.A, Monza, Italy). To evaluate the effect on A549 cells, 3 × 10 3 cells were seeded in 96-well plate. After adherence, cells were treated with Ni(SalPipNONO) (0.1 to 1 mM, 24 h) in presence of 0.1% e 2% FBS. To assay the effect on non-tumor cells on a condition of quiescence (to mimic a physiologic environment) 5 × 10 3 cells were seeded in 96-well plate. After adherence, cells were treated with Ni(SalPipNONO) at 0.5 and 1 mM for 24 h in presence of 0.1% FBS. Finally, to test the antiproliferative action of drug on non-tumor cells, 3 × 10 3 cells were seeded in 96-well plate. After adherence, cells were treated with Ni(SalPipNONO) at 0.5 and 1 mM for 24 h in presence of 2%. FBS. In all experimental conditions BrdU was added for the last 8 h of incubation. Then, cells were processed following manufacturer's protocol. Chemiluminescence generated by BrdU labelled cells was measured using Infinite F200 Pro (Tecan Life Sciences, Switzerland).

Clonogenic assay
The potential cytotoxic effect of NO donors was evaluated by the clonogenic assay [59]. A549 (2 × 10 5 cells/well) were seeded in 6-well plates. After adherence, cells were treated with Ni(SalPipNONO) or DETA/NO (0.1 to 1 mM, 48 h). Then, cells were trypsinized, seeded in 24 multi-well plate at the density of 500 cells/well and incubated at 37° C for 10 days in medium with 1% serum. Colonies were fixed with methanol and stained with a solution of Cristal Violet in 10% methanol (Sigma Aldrich, St. Louis, MO, USA). Colonies formed by over 30 cells were counted and representative pictures were shown. Data are expressed as surviving fraction.

ROS measurement
ROS levels were evaluated as previously reported [60]. A total of 1.5 × 10 3 cells were seeded in 96-well plates and, after adherence, were treated with 0.5 mM Ni(SalPipNONO) or Ni(SalPip) in medium without phenol red and different serum concentrations. NAC (5 mM, 30 min pretreatment) was used as a ROS scavenger. DCFH2-DA (2,-7-dichlorodihydrofluorescein diacetate; Invitrogen, Milan, Italy) was added (10 μM, 30 min) and intracellular levels of ROS were evaluated with a microplate reader (excitation/emission 495/527; Infinite F200 Pro (TecanLifeSciences, Switzerland). The results are reported as relative fluorescence units (RFU) corrected for the cell number counted.

Endothelial survival assay
Survival of endothelial cells (HUVEC) was evaluated following the protocol previously reported [61]. 1 × 10 3 cells/well (of 96-well multiplates) were let to adhere in 10% serum for 3-4 h and then VEGF (20 ng/ml) in presence/absence of the NO donor was added in medium with 0.1% serum. After 2 days, cells were fixed, stained and randomly counted at 20 × original magnification in 5 fields. Data are reported as number of cell counted/well.

In vitro co-culture assay
Tumor cells (3 × 10 4 cells) were cultured on transwell inserts (12 mm diameter, polycarbonate membranes with 0.4 µm pores; Corning, Lowell, MA, USA) and treated for 24 h with 0.5 mM Ni(SalPipNONO). Then the inserts were transferred on top of endothelial cells plated on Matrigel (1.5 × 10 5 cells in 12-well multiplate) for further 18 h of incubation. At the end of the experiment, endothelial cells were photographed and network formation on Matrigel was measured by means of the number of complete circles (Nikon Eclipse E400 and camera Nikon DS-5MC). www.impactjournals.com/oncotarget

In vivo matrigel angiogenesis assay
Investigation has been conducted in accordance with the ethical standards and according to the Declaration of Helsinki and the Italian law (Legislative Decree no. 26,4 March 2014), which acknowledges the European Directive 2010/63/UE, being approved by the authors' institutional review board and the Italian Ministry of Health. All efforts were made to minimize the number of animals used and their suffering. In vivo Matrigel angiogenesis assay was performed as previously described [32]. C57 black mice (20-25 g) were kept in temperature-and humiditycontrolled rooms (at 22° C) with lights on from 7 am to 7 pm, water and food available ad libitum. VEGF (300 ng) in presence/absence of Ni(SalPipNONO) (0.5 mM) was diluted in growth factor and phenol red-free Matrigel

Statistical analysis
Data represent means ± SD of at least 3 determinations. Statistical analysis was performed by means of Student's t test for unpaired data or by analysis of variance, followed by Bonferroni's test for comparison among groups of data; p < 0.05 was considered statistically significant.

ACKNOWLEDGMENTS
This work was partially funded by MIUR-PRIN project n. 2015Y3C5KP to LM.