Nitric Oxide Generated by Tumor-Associated Macrophages Is Responsible for Cancer Resistance to Cisplatin and Correlated With Syntaxin 4 and Acid Sphingomyelinase Inhibition

Tumor microenvironment is fundamental for cancer progression and chemoresistance. Among stromal cells tumor-associated macrophages (TAMs) represent the largest population of infiltrating inflammatory cells in malignant tumors, promoting their growth, invasion, and immune evasion. M2-polarized TAMs are endowed with the nitric oxide (NO)-generating enzyme inducible nitric oxide synthase (iNOS). NO has divergent effects on tumors, since it can either stimulate tumor cells growth or promote their death depending on the source of it; likewise the role of iNOS in cancer differs depending on the cell type. The role of NO generated by TAMs has not been investigated. Using different tumor models in vitro and in vivo we found that NO generated by iNOS of M2-polarized TAMs is able to protect tumor cells from apoptosis induced by the chemotherapeutic agent cisplatin (CDDP). Here, we demonstrate that the protective effect of NO depends on the inhibition of acid sphingomyelinase (A-SMase), which is activated by CDDP in a pathway involving the death receptor CD95. Mechanistic insights indicate that NO actions occur via generation of cyclic GMP and activation of protein kinase G (PKG), inducing phosphorylation of syntaxin 4 (synt4), a SNARE protein responsible for A-SMase trafficking and activation. Noteworthy, phosphorylation of synt4 at serine 78 by PKG is responsible for the proteasome-dependent degradation of synt4, which limits the CDDP-induced exposure of A-SMase to the plasma membrane of tumor cells. This inhibits the cytotoxic mechanism of CDDP reducing A-SMase-triggered apoptosis. This is the first demonstration that endogenous NO system is a key mechanism through which TAMs protect tumor cells from chemotherapeutic drug-induced apoptosis. The identification of the pathway responsible for A-SMase activity downregulation in tumors leading to chemoresistance warrants further investigations as a means to identify new anti-cancer molecules capable of specifically inhibiting synt4 degradation.

Tumor microenvironment is fundamental for cancer progression and chemoresistance. Among stromal cells tumor-associated macrophages (TAMs) represent the largest population of infiltrating inflammatory cells in malignant tumors, promoting their growth, invasion, and immune evasion. M2-polarized TAMs are endowed with the nitric oxide (NO)-generating enzyme inducible nitric oxide synthase (iNOS). NO has divergent effects on tumors, since it can either stimulate tumor cells growth or promote their death depending on the source of it; likewise the role of iNOS in cancer differs depending on the cell type. The role of NO generated by TAMs has not been investigated. Using different tumor models in vitro and in vivo we found that NO generated by iNOS of M2-polarized TAMs is able to protect tumor cells from apoptosis induced by the chemotherapeutic agent cisplatin (CDDP). Here, we demonstrate that the protective effect of NO depends on the inhibition of acid sphingomyelinase (A-SMase), which is activated by CDDP in a pathway involving the death receptor CD95. Mechanistic insights indicate that NO actions occur via generation of cyclic GMP and activation of protein kinase G (PKG), inducing phosphorylation of syntaxin 4 (synt4), a SNARE protein responsible for A-SMase trafficking and activation. Noteworthy, phosphorylation of synt4 at serine 78 by PKG is responsible for the proteasome-dependent degradation of synt4, which limits the CDDP-induced exposure of A-SMase to the plasma membrane of tumor cells.
inTrODUcTiOn Chemotherapeutic agent cisplatin (CDDP) is a widely used and very effective chemotherapeutic drug that induces apoptosis of cancer cells in solid tumors (1). However, CDDP has a major drawback; while patients in the first line of treatment usually respond to it, tumors develop resistance over time (2,3), mostly because of their ability to escape the apoptogenic effects of the drug (4). Recent studies have revealed a more complex scenario in which tumor resistance is also acquired through changes in the tumor milieu both in the cells composing the tumor microenvironment, as for instance immune cells, and in the soluble factors they release locally (5,6).
Inside the tumor microenvironment, tumor-associated macrophages (TAMs) have been demonstrated to be fundamental for cancer progression (7). In the tumor mass, these cells are subjected to a variety of stimuli that change their features. Initially, TAMs are in the M1 (classically activated) state, that is proinflammatory and anti-tumorigenic; as tumors progress they evolve to the M2 (alternatively activated) state, a pro-tumoral phenotype that triggers tissue remodeling as well as immune-suppression (8)(9)(10). TAMs are predominantly polarized as M2 macrophages in tumor of high grade associated with poor prognosis, and promote proliferation, survival, and motility of cancer cells (8,(10)(11)(12)(13)(14)(15)(16)(17).
Tumor-associated macrophages are competent to express inducible nitric oxide (NO) synthase (iNOS) and generate the gaseous messenger NO (18)(19)(20). At low, physiological levels NO displays cytoprotective properties while it is cytotoxic when produced at high concentrations (19,(21)(22)(23)(24). NO cytoprotection in cancer has been linked to the inhibition of the sphingomyelin metabolizing enzyme acid sphingomyelinase (A-SMase) whose activation is triggered by death receptors, i.e., CD95 and TNFRI, and chemotherapeutic drugs such as CDDP (25)(26)(27). We have previously found that an important determinant of A-SMase activity is the target-SNARE protein syntaxin 4 (synt4), known to be involved in several exocytosis events, such as antibody secretion as well as secretory pathway of CTLs and GLUT4 transport on the plasma membrane (28)(29)(30)(31). In U373 human glioma cells stimulated by CD95, we showed that synt4 presides over the vesicular trafficking of A-SMase allowing its exposure to the plasma membrane and hence its activation (28,32). Intriguingly syntaxins expression and activity, including that of synt4, have been linked to tumor progression through mechanisms still unknown (33)(34)(35)(36)(37).
We thus decided to investigate the molecular relationship between NO, A-SMase and synt4 and the role they play in the development of tumor resistance to CDDP. By in vivo and in vitro experiments using human and murine cell models we found that NO, produced at low levels by iNOS from M2-like TAMs, protects cancer cells from CDDP-induced apoptosis leading to chemoresistance. NO phosphorylates synt4 promoting its proteasomal degradation in a pathway involving generation of cyclic GMP (cGMP) and activation of protein kinase G (PKG). Synt4 downregulation in turn inhibits CDDP-induced A-SMase traslocation to the plasma membrane and its activation, thus blocking CDDP-apoptogenic action against tumor cells. We thus define for the first time a mechanism whereby NO regulation of synt4 causes chemoresistance to CDDP leading to the control of sphingolipid metabolism. The identification of this system in the tumor microenvironment contributes to shed light on the role of TAMs in the pathophysiology of cancer.

Viability assay
Cell viability of human and mouse glioma cells in the presence of CDDP at increasing concentrations for 24 h was evaluated by MTT analysis (40,45,46). MTT (Sigma-Aldrich, Saint Louis, MO, USA) absorbance was quantified spectrophotometrically using a Glomax Multi Detection System microplate reader (Promega, Madison, WI, USA).

Proliferation assay
U373 and GL261 cell proliferation after 24 h of treatment with CDDP was assessed by measuring the serial halving of cell fluorescence intensity via flow cytometry (45). The CytoTrack Cell Proliferation Assays (CytoTrack Green; Bio-Rad, Hercules, CA, USA) was used, according to the manufacturer's protocol. Fluorescence was analyzed by Gallios Flow Cytometer (Beckman-Coulter, Brea, CA, USA) and the software FCS Express 4 (De Novo System, Portland, OR, USA). The proliferation index, defined as the average number of cells that an initial cell became, was calculated using FCS Express software.

griess reaction
Nitric oxide generation in cell culture supernatants of human and murine M2 macrophages and by DETA-NO (20 µM) was assessed by the Griess method to measure nitrites, which are stable breakdown products of NO. Briefly, culture supernatants and DETA-NO were incubated with the Griess reagents I (1% sulphanilamide in 2.5% phosphoric acid) and II (0.1% naphthylenediamine in 2.5% phosphoric acid). The absorbance was read within 5 min at 550 nm and actual concentration calculated using a standard curve with serial dilutions of sodium nitrite.
Nine days after orthotopic implantation of tumor cells into the brain, mice were randomly assigned to receive the following treatments at days 9, 11, and 13: (a) intraperitoneal (i.p.) injection of PBS (control group), (b) i.p. injection of CDDP at 4 mg/kg (CDDP group); (c) i.p. injection of l-NAME at 4 mg/kg (l-NAME group); (d) i.p. injection of CDDP at 4 mg/kg combined with l-NAME at 4 mg/kg (l-NAME + CDDP group) ( Figure  S4A in Supplementary Material). Mice were sacrificed when they reached IACUC euthanasia criteria, as for instance clinical signs of tumor.
Intracranial tumor growth was monitored at days 9, 14, and 18 using the IVIS Lumina (Xenogen, Caliper Life Science-Perkin Elmer, Waltham, MS, USA). Mice were injected i.p. with d-luciferin (Beetle Luciferin Potassium Salt; Promega, Madison, WI, USA) at 50 mg/kg and then they were placed in the lighttight chamber, and a gray-scale image of the animals was first taken with dimmed light (FOV-Field of View-12.5 cm). After biodistribution time (20 min) photon emission was acquired for 10 min. Mice were acquired before and after each treatment. For co-localization of the bioluminescent photon emission on the animal body, gray-scale and pseudocolor images were merged using the Living Image Software ® (Caliper Life Sciences). To compare different mice, all images were scaled after all the acquisitions with the same scale and ROI analysis were performed to quantify luciferase signal intensity within tumor.

subcutaneous Tumor Model and Treatment
Lewis lung carcinoma (2 × 10 5 ) cells resuspended in PBS (Euroclone, Milan, Italy) were injected into the right flank of mice and allowed to grow. Nine days after tumor cells injection, mice were randomly assigned to receive the following treatments at days 9, 11, and 13: (a) i.p. injection of PBS (control group), (b) a daily i.p. injection of CDDP at 4 mg/kg (CDDP group); (c) i.p. injection of l-NAME at 4 mg/kg (l-NAME group); (d) i.p. injection of CDDP at 4 mg/kg combined with l-NAME at 4 mg/kg (l-NAME + CDDP group) ( Figure S4A in Supplementary Material). Tumor growth was monitored every 2-3 days by means of external caliper measurements and volume calculation (length × width 2 /2), until mice reached IACUC euthanasia criteria, as for instance clinical signs of tumor or when tumor size exceeded 10% of body weight (ca. 1,500 mm 3 tumor volume). When indicated, mice were sacrificed when tumor size reached ca. 500 mm 3 volume, and tumor collected for further analysis (i.e., TUNEL assay and western blotting).

TUnel assay
Lewis lung carcinoma tumors were collected and fixed in ice cold 4% paraformaldehyde before being rinsed in PBS and cryoprotected overnight in 30% sucrose. Tissues were then embedded in O.C.T. Compound (Sakura, AJAlphen aan den Rijn, The Netherlands) and cut in a CM1850 UV cryostat (Leica Biosystems, Wetzlar, Germany). At least five cryosections (6 µm) were assayed for apoptosis by the TUNEL method (DeadEnd Fluorometric TUNEL System), according to the manufacturer's protocol (46,47). Samples were counterstained with DAPI, mounted with Vectashield (Vector Laboratories, Burlingame, CA, USA), and examined using a DMI4000 B automated inverted microscope equipped with a DCF310 digital camera (Leica Microsystems, Wetzlar, Germany). Image acquisition was controlled by the Leica LAS AF software.

intratumoral Macrophage reconstitution and Tumor Treatment
Macrophage depletion was accomplished with injection of clodronate liposome (48). Briefly, 3 days after tumor injection, mice received a first i.p. administration of 200 µl of PBS liposomes (control) (PBS-LIPO) or clodronate liposomes (CL-LIPO; 1 mg/ml). Mice were, then, treated every 2 days with an intratumor injection of 50 µl PBS liposomes (control) or clodronate liposomes (0.25 mg/ml) (Liposoma B.V., Amsterdam, The Netherlands) till day 9 after tumor inoculation when macrophage depletion was assessed in two mice ( Figure  S4C in Supplementary Material). For macrophage reconstitution, 5 × 10 5 bone marrow-derived macrophages from wt or iNOS −/− mice were injected intratumor at day 11 and 13 after tumor inoculation. CDDP was i.p. administered at 4 mg/kg every 2 days from day 12 to 16 after tumor injection. Tumor growth was monitored every 2-3 days by means of external caliper measurements and volume calculation (length × width 2 /2), until mice reached IACUC euthanasia criteria, as for instance clinical signs of tumor or when tumor size exceeded 10% of body weight (ca. 1,500 mm 3 tumor volume).
All transformants were sequenced to verify the integrity of mutations. Transfection of U373 cells (~80% confluence) with the native and the mutated forms of synt4 was carried out using Fugene Tranfection Reagent (Promega, Madison, WI, USA), according the manufacturer's protocol. Analysis of synt4 expression levels was performed by western blotting 48 h after transfection.   resUlTs

TaMs induce chemoresistance of cancer to cDDP Through the generation of nO
The presence of M2-polarized macrophages in human glioma was assessed using sections of bioptic specimens of glioma, which were co-immunostained with the M2 subtype TAM marker CD206 and iNOS. Inside tumors we found areas with high density of cells positive for both proteins, namely M2-TAMs, and also cells positive only for iNOS, namely M1-TAMs ( Figure 1A). The presence of double positive cells demonstrates that M2-TAMs in gliomas are able to produce NO.
To investigate the role of NO generated by macrophages on CDDP-induced apoptosis we chose human U373 and mouse GL261 glioma cell lines, which are defective in the expression of both constitutive (nNOS and eNOS) and iNOS enzymes (Figures S1A-C and S2A-C in Supplementary Material). As shown in the viability assay of Figure 1B, U373 cells responded to CDDP (24 h) with an IC50 of 38.96 ± 0.17 µg/ml, a concentration at which cell proliferation was also significantly reduced ( Figure 1C). Glioma cells were co-cultured with human peripheral blood mononuclear cell-derived macrophages, polarized to M2 ( Figure S3A in Supplementary Material) and expressing detectable levels of iNOS ( Figure 1D). As shown in Figure 1E, the presence of macrophages protected glioma cells from CDDP (50 µg/ml)-induced apoptosis, an effect which was reversed in the presence of the NOS inhibitor l-NAME (2 mM, added to the cell culture 1 h before the administration of CDDP) (Figure 1E). The protective role of M2 macrophages on glioma cells treated with CDDP and l-NAME was restored after NO was re-added to the co-cultures using the NO donor DETA-NO (20 µM, 1 h before the administration of CDDP) (Figure 1E), that released a constant physiological flux of NO comparable to that of M2 macrophages (20, 54) ( Figure S3B in Supplementary Material).
Similar results were obtained with the GL261 cells. Their sensitivity to the apoptogenic action of CDDP (24 h) had an IC50 of 6.41 ± 0.17 µg/ml in the viability assay, a concentration that also significantly decreased cell proliferation (Figures 2A,B). GL261 cells were then co-cultured with M2-polarized macrophages derived from the bone marrow of wt mice (Figures S3C,D in Supplementary Material), that expressed detectable levels of iNOS ( Figure 2C). The apoptotic cell death of GL261 cells induced by CDDP treatment (20 µg/ml) was significantly reduced in the presence of wt M2 macrophages (Figures 2D,E). Notably, the presence of l-NAME restored CDDP effects but not when cells were also pre-incubated with DETA-NO ( Figure 2D). Taken together these findings indicate that NO produced by M2 macrophages protects tumor cells from CDDP-induced apoptosis. Accordingly, as shown in Figure 2E, the co-colture of GL261 cells with M2-polarized macrophages derived from the bone marrow of iNOS −/− mice (Figures S3C,D in Supplementary Material) did not modify the response to CDDP of glioma cells, further confirming the key role of M2-released NO. As expected, under the same experimental conditions we observed that CDDP induced apoptosis also in M2 macrophages. However, the cytotoxic effect of CDDP was different, being iNOS −/− cells significantly more sensitive when compared to wt cells (Figure 2F), thus suggesting an autocrine protective effect of NO.
To investigate the role of NO in vivo, we carried out experiments with the syngeneic tumor mouse models generated using GL261 glioma and LLC cell lines, also these latter being defective in the expression of NOSs ( Figures S2A,C in Supplementary Material), such that any effect of NO can be safely attributed to that generated in the host. GL261 (2 × 10 5 cells) and LLC (2 × 10 5 cells) were injected intraparenchymally (orthotopic allograft) and subcutaneously (flank allografts), respectively, in mice before i.p. administration with CDDP (4 mg/kg) both alone or in combination with l-NAME (3 mg/ml in drinking water) ( Figure S4A in Supplementary Material). By monitoring tumor progression over time we found that the largest reduction of tumor growth occurred in the CDDP + l-NAME-treated group (Figures 3A,B). In addition, TUNEL assay for apoptosis detection performed on LLC tumor slices demonstrated that CDDP-induced cell death was enhanced by l-NAME administration and the ensuing inhibition of NO generation (Figure 3C).  Figure S4A in Supplementary Material. Tumor growth was monitored by luminescence every 3 days. Panel on the left shows the bioluminescence imaging of a representative mouse for each experimental group at the indicated time points. The graph on the right shows the graphical representation of the bioluminescence expressed as fold induction compared to acquisition at day 9. (B,c) C57BL/6 mice (n = 7) were injected in the right flank with Lewis lung carcinoma (LLC) cells. Mice were then treated according to the scheme depicted in Figure S4A in Supplementary Material. (B) Tumor growth was monitored by measuring tumor volume (mm 3 ) every 2-3 days. *p < 0.05, **p < 0.01, and ***p < 0.001 vs untreated control mice. The bottom panel shows the typical photographs of subcutaneous LLC allografts excised from mice at day 25 of treatment. (c) Representative fluorescence micrographs of TUNEL and DAPI staining of LCC excided from mice when tumor size reached ca. 500 mm 3 (scale bar: 100 µm). (D) C57BL/6 mice (n = 7) were injected in the right flank with LLC cells. Mice were then treated according to the scheme depicted in Figure S4C in Supplementary Material. Tumor growth was monitored by measuring tumor volume (mm 3 ) every 2-3 days. *p < 0.05, **p < 0.01, and ***p < 0.001 vs CL-LIPO + wild-type MΦ + CDDP group. These data suggest cytoprotection as a mechanism by which NO limits the response to CDDP thus promoting tumor growth.
To assess whether macrophages in the tumor microenvironment generated the cytoprotective NO, we depleted TAMs in LLC masses by the injection of CL-LIPO in the peritumoral area (48,55) (Figure S4B in Supplementary Material). TAMs were subsequen tly replaced by macrophages obtained from wt and iNOS −/− mice inoculated inside LLC tumors before CDDP administration ( Figure S4C in Supplementary Material). As shown in Figure 3D, mice injected with iNOS −/− macrophages showed a significantly reduced tumor growth compared to mice injected with wt macrophages, indicating that NO released by TAMs is a key factor limiting tumor cell sensitivity to CDDP and thus identifying TAMs as a critical source of NO in the tumor microenvironment.
nO Mediates Tumor cells resistance to cDDP Through the generation of cgMP and the inhibition of a-sMase activity NOS/NO pathway induces many of its action via activation of guanylate cyclase/cGMP system (25,44,(56)(57)(58)(59). To corroborate our hypothesis that NO protects tumor cells from CDDP-induced apoptosis and to investigate the molecular mechanism underlying this event, U373 cells, natively deficient of NOS enzymes, were pre-incubated with the NO donor DETA-NO (20 µM, 1 h) before the addition of CDDP (50 µg/ml, 24 h). As shown in Figure 4A, Annexin V staining decreased in DETA-NO treated cells when compared to control (CDDP alone), thus revealing the paracrine key role of NO in inhibiting CDDP-induced apoptosis. Pre-incubation of U373 cells with 8Br-cGMP (3 mM, 1 h), a non-hydrolyzable and cell permeant analog of cGMP, mimicked the action of DETA-NO on CDDP-induced apoptosis while the protective effects of DETA-NO were blocked in the presence of ODQ (1 µM, administered 15 min before DETA-NO), a guanylate cyclase inhibitor that prevents NO-dependent cGMP generation (60,61). These results indicate that NO inhibits CDDP-induced apoptosis in tumor cells through the generation of cGMP.
The death receptor CD95 contributes to CDDP-induced apop tosis in cancer cells (32). In U373 cells, the pro-apoptotic effects of CDDP were inhibited, at least in part, by the presence of the neutralizing anti-CD95 antibody ZB4 (500 ng/ml, 1 h of pre-incubation) (Figure 4B), thus supporting the critical role of CD95 in our system. Similar results were obtained by brightfield microscopy revealing that the majority of cells exposed to CDDP alone, but not to ZB4 + CDDP, presented apoptotic cell morphological changes (round in shape, shrunken cytoplasm, formation of apoptotic bodies) (Figure 4C).
Acid sphingomyelinase, a valuable enzyme in cancer progression and in the sensitivity to chemotherapy (40,46) has been previously reported to be activated by CDDP (28,32,62). Therefore, we first assessed the time course of A-SMase activation by CDDP in U373 cells. Our results indicated that A-SMase activity transiently peaked at 30 min, returning to basal level at 60 min, and peaked again at 120 min after CDDP administration ( Figure 4D). One of the way through which NO protects cancer and normal cells from apoptosis induced by different stimuli, including CDDP, is through the modulation of A-SMase (25,26,57). Since NO protective effects against apoptosis may occur within few minutes after death stimuli (56), we analyzed the effect of NO during the first phase of CDDP-induced A-SMase activation. Pre-incubation of U373 cells with DETA-NO and 8Br-cGMP strongly inhibited A-SMase activity following 30 min of CDDP stimulation, that was instead maintained if DETA-NO was added to the cells in the presence of ODQ ( Figure 4E).
Commonly, A-SMase activation occurs by its translocation from cytosolic compartments to the plasma membrane (28,62,63). In U373 cells, cell surface biotinylation assay revealed that A-SMase specifically expressed at plasma membrane increased after 30 min of CDDP administration while this effect was reversed in the presence of DETA-NO and 8Br-cGMP ( Figure 4F). Of notice, no changes were detected in the total levels (mRNA and protein) of the enzyme in cells treated with CDDP, DETA-NO, and 8Br-cGMP both alone or in combinations ( Figures S5A,B in Supplementary Material). Taken together this data indicate that NO/cGMP inhibits A-SMase activation by preventing its translocation to the cell surface.

nO released by TaMs reduces synt4 expression in Tumor cells
The SNARE protein Synt4 controls the trafficking of A-SMase from the intracellular compartments to the plasma membrane and its activation upon CD95 stimulation (28) and NO/cGMP is known to regulate protein expression at both transcriptional and post-transcriptional levels (64)(65)(66). Treatment of U373 and GL261 cells with either 8Br-cGMP (3 mM) or DETA-NO (20 µM) for 1 h did not change the mRNA levels of synt4 (Figures 5A,B), thus excluding an effect at transcriptional level, still the protein amount was decreased significantly (Figures 5C,D). Likewise in co-cultures of GL261 with M2-polarized macrophages derived from the bone marrow of wt and iNOS −/− mice we found that the presence of wt M2 macrophages significantly reduced synt4 in GL261 cells when compared with control cells (GL261 alone) and with GL261 cells co-cultured with iNOS −/− M2 macrophages (Figure 5E). We sought independent evidence for this effect in vivo in LLC flank allografts implanted in wt and iNOS −/− mice, and analyzed 20 days after the tumor injection. In tumors extracts of iNOS −/− mice we found higher levels of synt4 compared to those observed in tumors implanted . Panel on the right shows apoptosis quantification expressed as fold increase of total apoptotic cells (annexin V + /PI − and annexin V + /PI + cells) compared to their respective untreated (UT) controls (dashed line) (n = 4). **p < 0.001 vs UT; ++ p < 0.05, ++ p < 0.001 vs CDDP. (B,c) U373 cells cultured in the presence of CDDP (50 µg/ml, 24 h) administered alone or together with neutralizing anti-CD95 antibody ZB4 (500 ng/ml, 1 h before CDDP). (B) Evaluation of CDDP-induced apoptosis expressed as fold increase of total apoptotic cells (annexin V + /PI − and annexin V + /PI + cells) compared to their respective UT (dashed line) (n = 3). *p < 0.05, **p < 0.001 vs UT; + p < 0.05 vs CDDP. (c) Brightfield microscopy images representative of three independent experiments. Scale bar: 100 µm. Bottom panels represent enlarged image details marked by the white arrows. (D) A-SMase activity on cell lysates measured as sphingomyelin hydrolysis to phosphorylcholine at pH 5.5. Cells (n = 3) were treated for the indicated time points (30,60, and 120 min) with CDDP (50 μg/ml). Enzyme activity is expressed as fold increase compared to UT (dashed line). *p < 0.05, **p < 0.001 vs UT. (e) A-SMase activity on cell lysates derived from cells cultured in the presence of CDDP (50 μg/ml, 30 min) administered alone or together with DETA-NO (20 µM, 1 h before CDDP), 8Br-cGMP (3 mM, 1 h before CDDP), and DETA-NO + ODQ (1 µM, 15 min before DETA-NO) (n = 3). Enzyme activity is expressed as fold increase compared to UT controls (dashed line). **p < 0.001 vs UT; ++ p < 0.001 vs CDDP. (F) A-SMase translocation was evaluated, at the indicated time points after CDDP (50 μg/ml) administered alone or together with DETA-NO (20 µM, 1 h before CDDP), 8Br-cGMP (3 mM, 1 h before CDDP), and DETA-NO + ODQ (1 µM, 15 min before DETA-NO), by assessment of biotinylated plasma membrane A-SMase, using the A-SMase antibody and assessing cytosolic GAPDH expression in parallel as internal control. The images are representative of three independent experiments. in wt mice ( Figure 5F). Together, all these data indicate that macrophage-derived NO decreases synt4 in tumors, at posttranscriptional level and in a pathway involving cGMP.

nO/cgMP induces Degradation of synt4 by the Proteasome via PKg-Dependent Phosphorylation of ser-78 Which explains the chemoresistance of Tumor cells
We then explored whether the post-translational reduction of synt4 by NO/cGMP was due to an action on the proteasomal pathway. U373 cells were treated with 8Br-cGMP (3 mM) for 1 h in the absence/presence of the proteasome inhibitor MG132 (10 µM). As shown in Figure 6A, MG132 reversed the degradation of synt4 induced by 8Br-cGMP. We also found that synt4 was ubiquitinated upon 8Br-cGMP treatment (Figure 6B), thus indicating that the reduction of synt4 expression was mediated by NO/cGMP through a proteasome-dependent degradation. The PKG, a classical signaling molecule activated by NO/ cGMP, is involved in the inhibition of A-SMase activity (25) and serine phosphorylation is known to regulate protein degradation (67)(68)(69)(70). We found that the effect of 8Br-cGMP on synt4 levels of U373 cells was inhibited in the presence of KT5823 (1 µM), a well-known PKG inhibitor ( Figure 6C). We thus analyzed the human and murine synt4 primary sequence using the NetPhos phosphorylation prediction algorithm, identifying potential PKG phosphorylation residues located either at the NH2-terminus regulatory domain (Ser-78) or in the t-SNARE domain (Ser-208, Ser-216, Ser-247). The serine residue of each site was mutated to alanine by site-directed mutagenesis. The effect of 8Br-cGMP on the degradation of the mutants, transiently overexpressed in U373 cells, was then evaluated. As shown in Figure 6D, the mutation of Ser-78 (S78A) conferred resistance to cGMPinduced synt4 degradation, while none of the other mutations had any significant effect, thus indicating that phosphorylation of Ser-78 site of synt4 protein is required for PKG action.
Finally, we investigated CDDP-induced A-SMase activity and apoptosis in U373 cells transfected with the synt4 S78A mutant. Cells expressing this mutant remained sensitive to CDDP (50 µg/ml) even in the presence of 8Br-cGMP; indeed, the administration of 8Br-cGMP did not alter the effect of CDDP in terms of A-SMase activity and apoptosis (Figures 6E,F). Taken together, our data indicate that synt4 phosphorylation at Ser-78 and its ensuing degradation by NO/cGMP/PKG is indeed responsible for the resistance of tumor cells to CDDP via A-SMase inhibition.

DiscUssiOn
In this study, we report a new mechanism at the basis of resistance of solid tumor of various origin to chemotherapeutics through an action on A-SMase. This event is based on NO produced by M2-like macrophages in the tumor microenvironment that decreases the sensitivity of tumoral cells to CDDP-induced apoptosis. We also identified in the degradation of the SNARE protein synt4 and in the ensuing inhibition of A-SMase plasma membrane exposure the molecular mechanism of NO action.
The effect of NO on tumor biology has been largely explored and has been demonstrated to be dichotomous, since it can either stimulate tumor cells growth or promote their death depending on its concentration and origin (21)(22)(23)(24)71). Administration of exogenous NO may directly kill cancer cells or act as chemosensitizing agents (72,73), thus indicating NO-donors as potential anticancer drugs (74,75). However, it has to be pointed out that the final activity of exogenous NO in cancer depends on its microenvironment, the type of cell exposed to the compound and its redox state, as well as the intracellular concentration and the duration of intracellular exposure to NO (74).
Another issue currently debated is the role of iNOS in cancer. Overexpression of iNOS and concomitant changes on cancer cell kinetics are demonstrated to have an anti-cancer action according to several in vitro and in vivo studies (71,(76)(77)(78). However, the observation that iNOS expression is high in a number of tumors and that this correlates with poor survival, has also led to the conclusion that induction of iNOS may somehow be related to tumorigenesis and tumor growth (79)(80)(81)(82).
In light of this, it appears that the dual role of iNOS is strongly influenced by the cell situation and depends on the environment, with either induction or inhibition of iNOS having anti-cancer or pro-cancer potential based on tumor and cell types.
Inside the tumor mass, NO can be produced by both cancer cells and cells of the tumor microenvironment among which  TAMs. In early events of tumorigenesis, TAMs show a M1-like phenotype, highly expressing iNOS that generates huge concentrations of NO, which in turn trigger tumor cell apoptosis and erase newly transformed cells (71). During tumor development, TAMs acquire a M2-like phenotype and promote tumor development possibly contributing also to resistance to chemotherapy   (83,84). It is worth mentioning that M2-like macrophages show a redu ced expression of iNOS and generate NO at low, cytoprotective concentration (20). By in vitro experiments carried out in human and murine tumor cell models, i.e. U373 and Gl261 glioma cells, that do not natively express NOSs, and culturing these cells with macrophages derived from the bone marrow of wt and iNOS −/− mice, we observed that the presence in tumor milieu of M2-like macrophages expressing iNOS renders tumor cells resistant to CDDP by protecting them from apoptosis. Moreover, analyzing cell death of wt and iNOS −/− derived M2 macrophages, we found that iNOS −/− cells were more sensitive to apoptosis, thus suggesting a possible paracrine/autocrine effect of NO (85). Inhibiting iNOS activity in vitro in both cell lines and in vivo in tumor allografts, by the administration of NOS inhibitor l-NAME increased the apoptotic activity of CDDP and reduced significantly tumor growth. The anti-proliferative effect of l-NAME has been demonstrated in different kinds of cancer cells in vivo (86)(87)(88). Our results obtained by the replacement of wt TAMs with iNOS −/− TAMs in tumor allografts indicate that NO generated by TAMs is a key player in CDDP resistance. Despite being one of the oldest chemotherapeutic drugs, CDDP is still a mainstay in the therapy of a variety of solid tumors (1). The efficacy of this drug is due to its multiple actions. Alongside its direct cytotoxic action at the DNA level, CDDP targets molecules promoting tumor expansion among which are cytoskeleton proteins, mitochondrial DNA, and plasma membrane proteins and lipids (2,3,62). CDDP activates also the CD95 death receptor on tumor cells and triggers A-SMase activity (26,32,62,89). This leads to the activation of downstream apoptotic pathways and to a modification of the plasma membrane structure such that receptor clustering is favored and the death signal amplified (26,32,62,90). Here, through an experiment of CD95 blockade, we corroborate the findings that the death receptor activation contributes, at least in part, to CDDP-induced cell death. The partial effect on the inhibition of apoptosis we observed was expected, because of the multiple ways through which CDDP may kill cancer cells (3).
We have previously reported that NO protects cells from apoptosis induced by different stress stimuli through the inhibition of A-SMase activity (25,26,56). Here, we demonstrated that such an action, at least in the case of CDDP signaling, is dependent on NO ability to generate cGMP in tumoral cells and to block the CDDP-induced, synt4-dependent A-SMase translocation on the plasma membrane and hence its activity (28). Furthermore, we demonstrated that this action of NO is mediated through synt4 phosphorylation by PKG on Ser-78 and this in turn triggers proteasome-dependent degradation of synt4. The phosphorylation of synt4 at Ser-78 site within the NH2terminal regulatory domain of the t-SNARE we describe here stimulates protein degradation, likely inducing the misfolding of synt4 that triggers its recognition by the proteasome.
The very tight relationship between proteasome regulation and CD95-induced apoptosis (91) suggests the importance in cell death of the pathway of synt4 phosphorylation/degradation we report. This pathway may also be important in view of the role played by NO and A-SMase in the pathogenesis of cardiovascular, pulmonary, liver, and neurological diseases, as well as sepsis and infections (57,(92)(93)(94)(95). Previous studies showed that synt4 can be phosphorylated by protein kinases A, Cα, and casein kinase II and that phosphorylation is a fundamental mechanism for synt4dependent exocytosis, since binding of synt4 to cognate SNAREs is altered by its phosphorylation (96). Synt4 phosphorylation by protein kinase Cα promotes exocytosis of Weibel-Palade bodies in endothelial cells by dissociating the synt4/Munc-18 complex (97), whereas phosphorylation by protein kinase A inhibits synt4 binding to SNAP 23 (96). Recently, a role for synt4 phosphorylation state has been demonstrated for the regulation of membrane type-1 matrix metalloproteinase trafficking during invadopodium formation and tumor cell invasion (36,37). Protein phosphorylation is known to regulate and maintain protein function and turnover, also by the activation of proteasome in physiological and pathological condition (67-70, 98, 99). Although the PKGdependent phosphorylation of synt4 and the ensuing degradation had not been reported yet, it is conceivable that such event characterizes also the above biological systems in which both A-SMase activity and NO generation occur. Finally, the phosphorylationinduced degradation of synt4 by proteasome as a novel mechanism inducing chemoresistance to CDDP may be considered relevant also in the light of the recent advance in combinatorial cancer therapy based on proteasome inhibitors (100)(101)(102).
In summary, we demonstrated here that NO produced by iNOS expressed in M2-like TAMs efficiently protects cancer cells from CDDP-induced apoptosis leading to chemoresistance, such that inhibition of iNOS in TAMs is per se sufficient to restore the efficacy of chemotherapy. In addition, we found that NO activates cGMP/PKG pathway in cancer cells leading to phosphorylation of synt4 at Ser-78 site, an event which promotes the proteasomal-dependent degradation of synt4. Importantly, low levels of synt4 limit the CDDP-induced exposure of A-SMase to the plasma membrane of tumor cells thus inhibiting the cytotoxic mechanism of this drug. The identification of the role of this pathway in chemoresistance of tumors warrants further investigations as a means to identify new anti-cancer molecules capable of specifically inhibiting synt4 degradation.

eThics sTaTeMenT
All studies involving animals were conducted in accordance with the Italian law on animal care N° 116/1992 and the European Communities Council Directive EEC/609/86. The experimental protocol was approved by the Ethics Committee of the Università degli Studi di Milano.
aUThOr cOnTriBUTiOns CP initiated the project, designed and supervised all the experiments, collected and processed the samples, analyzed the data, and wrote the article. DC and CDP processed the samples and analyzed the data, contributed to the experimental design, work discussion, and article writing. PR and MB contributed to the experimental design and to work discussion. LC and PR-Q performed the analyses on human macrophages. CM, LO, and GL performed the analyses on orthotopic brain tumor model. CM and ECa were responsible for immunofluorescence experiments. IR, MG, SZ, and MC contributed to sample collection, processing, and analysis of in vitro experiments. AC contributed to animal handling and in vivo experiments. EC coordinated the experimental work and contributed to article writing.

FUnDing
This work was supported by: "Università di Milano, Giovani Ricercatori-Linea 2" grant to CP; "Ministero dell'Istruzione, Università e Ricerca, " PRIN2015 (2015MJBEM2) grant to EC and DC. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.