Vascular Caveolin Deficiency Supports the Angiogenic Effects of Nitrite, a Major End Product of Nitric Oxide Metabolism in Tumors

Nitric oxide (NO) plays a major role in many key pathophysiologic processes, including the control of vascular tone (1) and angiogenesis (2). Although nitrites and nitrates have been described for a long time as the inert end products of the NO oxidative metabolism (3), recent evidence indicates that under specific conditions nitrite can be reconverted into biologically active NO (4). The nitrite anion is now considered as the largest intravascular and tissue storage of NO, which may be made available depending on the tissue need. Nitrite was for instance shown to contribute to hypoxic vasodilation (5-7) and to protect several organs, including heart, liver, kidney, and brain, from ischemia-reperfusion injuries (8-13). Different pathways are proposed to support the reductive reconversion of nitrite into NO. Besides the acidic reduction/disproportionation of nitrite species (14-16), some heme-containing enzymes, including deoxyhemoglobin (7, 17, 18) and xanthine oxidase (19, 20), can act as nitrite reductases under hypoxia, thereby offering a salvage pathway to produce NO when oxygen-consuming NO synthases (NOS) become inoperative.

Interestingly, the endothelial NOS (eNOS) isoform itself, in the absence of oxygen, has been shown to behave as a nitrite reductase (21, 22). In endothelial cells, the catalytic eNOS activity is regulated by caveolin-1, the structural protein of flask-shaped plasma membrane invaginations named caveolae. The physical interaction of eNOS with the so-called caveolin-1 scaffolding domain maintains the enzyme in an inactivated state (23-26), preventing the activation of eNOS in basal (unstimulated) conditions. In tumors, we recently reported that a deficit in the expression of caveolin in tumor blood vessels was associated with an exacerbated angiogenesis (27). Although caveolin-free eNOS as a direct source of NO (endowed with potent angiogenic properties) may account for the observed increase in tumor neovascularization, the low pO₂ in tumors should, however, alter the capacity of the enzyme to produce NO. Here, we examined the alternative hypothesis according to which nitrite reconverted into NO by caveolin-free eNOS in tumor blood vessels could support angiogenesis occurring under hypoxia. We used the so-called ex vivo aorta ring assay where pieces of freshly dissected mouse aorta are embedded in collagen and cultured in a medium complemented with serum.

We found that the lack of caveolin-1 expression leads to an extended sensitivity of vessels to nitrite exposure and consecutively favors NO-dependent angiogenesis under hypoxic conditions. As a decrease in caveolin-1 expression was reported in endothelial cells lining tumor blood vessels (27, 28), these findings underline the potential of nitrite in promoting angiogenesis in the hypoxic tumor microenvironment and strengthen the need for strategies targeting NO production in tumors to affect cancer progression.

To assess whether caveolin down-regulation, as observed in tumor blood vessels, could regulate angiogenesis, sera from Cav^+/+ and Cav^−/− mice were used to stimulate vascular outgrowth from aortic rings (themselves collected from Cav^+/+ and Cav^−/− mice). Aorta rings were embedded in collagen gels and cultured in the presence of autologous serum (i.e., Cav^+/+ serum for Cav^+/+ explants and Cav^−/− serum for Cav^−/− explants). As shown in Fig. 1A and B, the outgrowth of microvascular structures was more extensive from Cav^−/− aorta than from Cav^+/+ explants. To discriminate between the respective influence of the explants and the serum on the promotion of angiogenesis, Cav^+/+ and Cav^−/− aortic rings were also cultured in the presence of heterologous serum (i.e., Cav^−/− serum for Cav^+/+ explants and Cav^+/+ serum for Cav^−/− explants). As presented in Fig. 1A and C, Cav^−/− serum promoted outgrowth of microvascular structures from both genotype explants to a higher extent than wild-type serum. Cav^−/− serum did, however, induce the outgrowth of more endothelial structures from Cav^−/− than from Cav^+/+ explants.

To determine the nature of the proangiogenic factors present in the Cav^−/− mouse serum, we used protein array membranes designed to evaluate the abundance of major proangiogenic and antiangiogenic actors. In this assay, no critical differences were found in the composition of sera from Cav^+/+ and Cav^−/− mice (Fig. 2A). We then focused on the major difference between Cav^+/+ and Cav^−/− mouse vessels: the unclamping of eNOS from its inhibitory interaction with caveolin in endothelial cells and the consecutive increase in eNOS activity (23, 29). We found that, indeed, agonist-induced NO release from Cav^−/− aorta amounted to ∼160% of the NO production in wild-type vessels (Fig. 2B) and that a ∼3-fold increase in serum nitrite concentration was found in Cav^−/− mice (versus Cav^+/+ mice; Fig. 2C). The L-NAME sensitivity of the NO measurement confirmed that changes in eNOS activity did account for the observed differences (Fig. 2B); note that no change in eNOS protein abundance was observed according to the two mouse genotypes (data not shown).

As nitrite can reform biologically active NO (which in turn may exert proangiogenic effects), we then examined whether the treatment of aorta rings with nitrite could recapitulate the effects of Cav^−/− serum. As shown in Fig. 3A and B, addition of exogenous nitrite led to a dramatic increase in endothelial outgrowth from Cav^−/− aorta rings.

Based on the above results, we suspected that eNOS was acting as a nitrite reductase in our model. We therefore aimed to examine whether hypoxia (a prerequisite for such activity) was present in our experimental conditions. We evaluated the expression of carbonic anhydrase IX (CAIX) as a reporter of hypoxia in our models of embedded aorta rings (wherein oxygen gradient could develop through the collagen gel). Although experiments were done under normoxic conditions (5% CO₂/95% air in the incubator), the CAIX immunostaining was found to be higher when explants were embedded in collagen compared with explants freshly isolated or cultured but nonembedded (Fig. 4A and B). Aortic rings submitted to 1% hypoxia were used as positive controls for the induction of CAIX expression. Of note, electron paramagnetic resonance measurements using lithium phthalocyanine as an oxygen-sensitive paramagnetic probe also indicated that the pO₂ in collagen-embedded aorta was consistently found to be ∼15 mm Hg (despite technical limitations linked to the processing of the vessel out of the incubator).

To document the role of eNOS as a major nitrite reductase in our ex vivo assay, we used aorta rings from eNOS-deficient mice. As shown in Fig. 4C, we observed that nitrite failed to stimulate vascular outgrowth from eNOS^−/− aorta rings. We then measured the capacity of eNOS^−/−, Cav^−/−, and wild-type mouse aorta rings to convert l-arginine into NO. Such measurements of the catalytic eNOS activity led to the observation that collagen embedding largely prevented the NOS activity (see white columns in Fig. 4D). Interestingly, exposure of wild-type and Cav^−/− aorta rings to high O₂ concentrations led to a net stimulation of NOS activity in this assay (see black columns in Fig. 4D); the eNOS^−/− aorta rings completely failed to reveal NOS activity. Note also that the extent of l-arginine conversion was not significantly different between Cav^−/− and wild-type mouse aorta and that this oxidative NO-producing pathway could thus not account for the difference in the extent of angiogenesis observed between these two conditions (see Fig. 4C). We also exploited the capacity of endothelial cells to form tubes when cultured on Matrigel to evaluate whether the combination of hypoxia and exogenous nitrite could modulate angiogenesis. In this other assay, hypoxic conditions (1% O₂) failed to stimulate basal angiogenesis from wild-type endothelial cells (see hatched columns in Fig. 4E). We found, however, that the addition of 100 μmol/L nitrite significantly stimulated endothelial tubulogenesis, reaching 20% of the effects observed under normoxia. Caveolin-1 silencing using a dedicated small interfering RNA (siRNA) further increased by 2-fold the extent of tube formation from endothelial cells exposed to hypoxia in the presence of nitrite (Fig. 4E). Caveolin deficiency barely influences, in this model, the extent of basal angiogenesis under normoxia. Finally, using endothelial cells isolated from eNOS^−/− mice, we confirmed the key role of this enzyme in both normoxic and hypoxic conditions to drive angiogenesis because we did not observe any significant tube formation by plating theses cells on Matrigel (Fig. 4E). Moreover, caveolin silencing in these eNOS^−/− endothelial cells did not have any effect on the capacity of nitrite to stimulate angiogenesis under hypoxia (Fig. 4E).

To assess the NO dependence of both the Cav^−/− serum and the exogenous nitrite-induced angiogenesis in our ex vivo model, we exposed aorta ring cultures to 100 μmol/L NO scavenger (C-PTIO). We found that C-PTIO abrogated the stimulation of the vascular outgrowth observed in response to Cav^+/+ serum complemented with exogenous nitrite (100 μmol/L; see Fig. 5A and B). Importantly, C-PTIO also reduced by 60% the microvascular structure outgrowth from aorta rings exposed to Cav^−/− serum, confirming the NO dependency of the pathway induced by nitrite-rich serum (Fig. 5A and B); vehicle (DMSO) had no effect on the extent of endothelial outgrowth (data not shown).

We have compared the nitrite content in tumors grown in Cav^−/− and Cav^+/+ mice. We found a slight but reproducible 20% increase in Cav^−/− tumors (Fig. 6A). We also documented that the nitrite concentration was 5-fold higher in tumors than in the surrounding muscle (Fig. 6B). Using the specific inducible NO synthase (iNOS) inhibitor 1400W, we found that most of the tumor versus muscle difference arose from iNOS activity (Fig. 6B). Finally, using pools of tumors of small and large diameters, we collected tumors with different O₂ status, as determined by electronic paramagnetic resonance (EPR), and found that hypoxic tumors were associated with larger nitrite contents (Fig. 6C).

In this study, we found that the reduction in caveolin abundance known to occur in tumor vessels (27, 28, 30) is prone to favor NO-mediated angiogenesis from nitrite. The ex vivo model of serum-driven vascular outgrowth from aorta rings was critical to identify how nitrite could take advantage of a caveolin-unclamped eNOS to promote angiogenesis. The higher angiogenic potential of the serum collected from Cav^−/− mice (versus wild-type mice) was the seminal finding that led us to identify nitrite as key actor supporting vascular outgrowth. As a preamble to this discussion, it should, however, be stressed that the increase in serum nitrite concentration in Cav^−/− mice arises from the overall increase in eNOS activity in all the vascular beds of the animal and therefore does not reflect a naturally occurring systemic increase in nitrite in response to local tumor growth. The accumulation of nitrite, as a major end product of the NO catabolism, in tumors may actually arise from the activation of any NOS isoform. We and others have previously reported that the abundance of caveolin, the scaffold protein regulating eNOS activity (23, 29), was reduced in the tumor vasculature (27, 28, 30), thereby promoting the endothelial NO production in tumors. High levels of NO production may also result from iNOS-expressing tumor-associated macrophages (31, 32) or directly from neuronal NOS–expressing tumor cells (33); some tumor cells were also reported to express iNOS or eNOS (34). In the current study, using a mouse tumor model wherein tumor cells do not express any NOS isoform, we documented that the major contribution of host cells in the production of nitrites by the tumor arose from iNOS (Fig. 6).

The fate of nitrite is normally to be eliminated in the urine after transiting into the bloodstream. In ischemic cardiovascular tissues, it was, however, recently documented that infused nitrite could contribute to reform a bioactive NO pool and participate to the recovery of ischemic tissues (35, 36). Our data indicate that nitrite-derived NO may exert its effects by stimulating angiogenesis. This may participate in the reduction of postischemic damages in cardiovascular tissues but also in the development of tumor neovascularization. Among the enzyme candidates for the reconversion of nitrite into NO, our work on Cav^−/− mice indicates that eNOS may act as a nitrite reductase, particularly when the abundance of the endogenous inhibitor caveolin is reduced and under hypoxic conditions. The combination of these two factors is critical. Indeed, in vitro data previously documented that under complete anoxia, eNOS could contribute to reconvert nitrite into NO (21, 22). In our experimental model of aorta rings, hypoxia is present, as verified by CAIX staining and EPR oximetry probe, but should be viewed as a moderate form of hypoxia because it is the embedding of aorta rings into the collagen (and not the strict lack of O₂) that limits the O₂ diffusion to the vessel from the ambient incubator 95% air-5% CO₂ gas mixture. Importantly, we showed that this form of diffusion-limited hypoxia was sufficient to prevent significant conversion of l-arginine into NO (i.e., normal eNOS catalytic activity) and could be reversed by increasing oxygen concentration in the medium bathing embedded aorta rings (Fig. 4C).

We and others previously documented that tumor angiogenesis was exacerbated in tumor growing in Cav^−/− mice (27, 37). The current data support the contribution of NO—through nitrite reconversion—to the neovascularization process occurring in tumor hypoxic regions. Using cultured endothelial cells exposed to caveolin-targeting siRNA, we showed that although the extent of angiogenesis was lower than in normoxic conditions, the caveolin down-regulation promoted the angiogenic effects of the combination of exogenous nitrite and hypoxia (Fig. 4E).

In conclusion, our data identify the angiogenic potential of nitrite in tumors and unravel how eNOS-mediated signaling may occur in hostile, hypoxic conditions, as encountered in many tumors. This study also strengthens the therapeutic value of caveolin reexpression (38) or caveolin-derived peptide administration (39, 40) to selectively block eNOS activity in tumors.

Caveolin-1–deficient (Cav^−/−) mice (originally obtained from Dr. T.V. Kurzchalia, Max Planck Institute for Molecular Cell Biology and Genetics, Dresden, Germany; ref. 41) and their control littermates (Cav^+/+) were generated through heterozygous matings and housed in our local facility. eNOS-deficient (eNOS^−/−) mice were obtained from Charles River Laboratories. NMRI, male mice (Elevage Janvier) were used in experiments with syngeneic TLT carcinoma cells (42, 43). Mice received an i.m. injection of 10⁵ tumor cells in the posterior right leg. The tumor diameters were tracked with an electronic caliper. In some experiments, mice received the iNOS-selective inhibitor 1400W (50 mg/L in the drinking water). Each procedure was approved by the local authorities according to national animal care regulations.

Preparation of three-dimensional aortic ring cultures was made as previously described (44). Briefly, blood and aorta were collected from anesthetized mice. Cleaned aortic rings (1-mm long) were then embedded in a collagen type I matrix and cultured in the presence of MCDB131 medium supplemented with 25 mmol/L NaHCO₃, 2.5% mouse serum, 1% glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. In each experiment, mouse sera were pooled in accordance with genotype. Digital images of microvascular structure outgrowth were obtained with the use of a Zeiss inverted microscope coupled to a charge-coupled device camera on days 6, 8, 10, and 12. Angiogenesis quantification was done by counting the number of microvascular structures grown out of each aortic explant. Nitrite (NaNO₂; Fluka) and C-PTIO [2-(4-carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide potassium salt; Sigma] were added at the day of the culture preparation.

Endothelial cells were obtained from wild-type and eNOS^−/− mouse aorta according to the primary explant procedure, as previously reported (45). Endothelial cell reorganization in capillary-like structures was observed, after plating on Matrigel, using an inverted phase-contrast microscope and the length of the endothelial network was quantified by analysis of images randomly captured by a video camera system. In some experiments, endothelial cells were transfected with duplex siRNA targeting caveolin-1 (corresponding to the coding sequence 206-226) using Lipofectin (Invitrogen), according to the manufacturer's protocol. A negative control siRNA was used as control and a minimum 90% reduction in caveolin-1 expression level was validated by immunoblotting done in parallel to the assay, as described elsewhere (27). For hypoxic conditions, cells were incubated in a gas-regulated (1% O₂, 94% N₂, 5% CO₂) workstation (InVivo 500, Ruskinn).

Mouse Angiogenesis Antibody Array I (RayBiotech, Inc.) was used to compare the relative amounts of different angiogenesis regulator proteins present in the sera of Cav^−/− and Cav^+/+ mice. Sera were collected from mice just as for the three-dimensional aortic ring cultures. Blood sera from three animals per genotype were pooled and screened for cytokine expression, according to the manufacturer's protocol. Quantification was done by densitometry using ImageJ 1.37v (NIH).

After 2 d of culture, aortic rings were drawn from the collagen gels with fine forceps and quickly frozen in liquid nitrogen for cryoslicing. Rings were also either directly snap frozen or cultured (nonembedded) in complete MCDB131 medium for 2 d. In some experiments, embedded aortic rings were submitted to hypoxia (1% O₂) for 18 h in a modular incubator chamber (Billups Rothenberg, Inc.) used as previously described (43). Green autofluorescence of elastic fibers in the aorta wall was quenched using Chicago Sky Blue 6B 0.5% (Sigma), and sections were then probed with a goat antibody against CAIX (R&D Systems) and a secondary anti-goat Alexa Fluor 488–coupled antibody (green fluorescence; Invitrogen). Nuclei were stained using 4′,6-diamidino-2-phenylindole (Sigma). Slices were examined with an Axioskop microscope (Zeiss) equipped for fluorescence, pictures were acquired using a charge-coupled device camera, and quantification of the signal intensity was done using ImageJ 1.37v.

Thoracic aorta were microdissected, placed in ice-cold physiologic solution [Krebs solution with 10 mmol/L Na-HEPES (pH 7.6)], and cleaned of all fat and connective tissue. The NO production was assayed by EPR spin trapping as described previously after formation of the paramagnetic adduct Fe(II)NO(diethyldithiocarbamate)₂ {[Fe(II)NO(DETC)₂]} in tissue treated for 30 min at 37°C with 0.4 mmol/L [Fe(II)(DETC)₂] complex as colloid and stimulated (or not) with ionomycin (2 μmol/L). For NO measurements by spin trapping, tissues were then rapidly frozen in calibrated tubes (0.3 mL) and kept in liquid nitrogen for EPR measurements. The EPR spectra were recorded with an X-band EMX at 77 K using a finger Dewar vial. A NO colorimetric assay (Roche Diagnostics) was used to determine the accumulation of NO derivatives (nitrite + nitrate) in the mouse plasma, muscle, or tumor. The activity of eNOS was also determined from embedded aorta rings by measuring the conversion of [³H]arginine to [³H]citrulline, as previously described (46). During the incubation period, embedded aortic rings were either maintained under normoxic conditions (21% O₂) or exposed to 95% O₂-5% CO₂ gas mixture in a modular incubator chamber.

EPR oximetry, using charcoal (CX0670-1, EM Science) or lithium phthalocyanine as oxygen-sensitive probes, was used to evaluate pO₂ levels in tumor and in collagen-embedded aorta, respectively, as previously described (42, 47).

Data are reported as mean ± SE, and statistical analyses were done using Student's t test or two-way ANOVA analyses where appropriate.

No potential conflicts of interest were disclosed.

We thank Giovanna Gisellu and Julie Dehareng for their excellent technical assistance.