Viagra releases the brakes on melanoma growth

ABSTRACT

Use of the phosphodiesterase 5 (PDE5) blocker sildenafil (Viagra) has been linked to an increased risk of melanoma. In a recent study we discovered a growth-promoting cGMP pathway in melanoma cells. cGMP degradation via PDE5 acts as a brake on this pathway. Inhibition of PDE5 releases this brake, providing a mechanism for the promelanoma effects of sildenafil.

Malignant melanoma is one of the most aggressive cancers. At the molecular level, the mitogen-activated protein kinase (MAPK) pathway is hyperactivated in the majority of melanomas, typically through somatic gain-of-function mutations of the BRAF gene. In 2011, Richard Marais' group showed that oncogenic BRAF signaling and invasiveness of melanoma cells in mice are associated with an increased level of cyclic guanosine-3´, 5´-monophosphate (cGMP) as a result of transcriptional repression of the PDE5A gene, which encodes the cGMP-degrading phosphodiesterase 5 (PDE5).¹ This study raised concerns that PDE5 inhibitors such as sildenafil (Viagra) that increase intracellular cGMP concentrations may promote malignant melanoma. PDE5 inhibitors are used to treat erectile dysfunction, pulmonary hypertension, and an increasing number of other diseases. Indeed, 2 clinical studies reported that use of PDE5 inhibitors in men is linked to a modest increase in melanoma risk.^2,3 However, these studies did not prove causality of the observed associations and there is debate over whether the higher incidence of melanoma in men using PDE5 inhibitors was in fact caused by a biological effect of the drug on tumorigenesis or rather due to their lifestyle, which included higher levels of sun exposure.

In a recent study we identified the molecular players of cGMP signaling in murine and human melanoma cells.⁴ C-type natriuretic peptide (CNP), a ligand of the membrane-bound guanylate cyclase B (GC-B), strongly increased intracellular cGMP concentrations and the activity of cGMP-dependent protein kinase I (PRKG1, also known as PKG or cGKI) in tumor cells. Activation of this cGMP pathway promoted p44/42 MAPK signaling, melanoma cell growth, and migration in vitro. These effects were potentiated by sildenafil. Importantly, overexpression of cGKI in melanoma cells or administration of sildenafil enhanced melanoma growth in mice in vivo. Based on these findings, we propose a model (Fig. 1) in which CNP acts as fuel and PDE5 as a brake on the growth-promoting cGMP pathway. Sildenafil releases the PDE5 brake, thus potentiating activation of the cGMP pathway and promoting the switch of non-metastatic cells in primary melanomas to invasive/metastatic cells. Clearly, there are many important questions regarding cGMP signaling and the use of cGMP-elevating drugs in melanoma patients.

Figure 1.

cGMP, phosphodiesterase 5 (PDE5), sildenafil, and melanoma growth. Melanoma cells (gray) express a cGMP pathway (not shown) comprised of the CNP receptor guanylate cyclase B (GC-B), cGMP-dependent protein kinase I (cGKI), and the cGMP-degrading PDE5. The peptide hormone C-type natriuretic peptide (CNP) is released from inflamed tumor vessels (green jerrycan) and triggers cGMP synthesis in melanoma cells via GC-B. cGMP activates cGKI, which promotes mitogen-activated protein kinase (MAPK) signaling via phosphorylation of currently unknown substrate proteins. Cells with increased potential for growth, migration, and invasiveness develop. PDE5 hydrolyzes cGMP, keeping cGMP levels and melanoma growth low (left). Thus, CNP acts as fuel (left and right) and PDE5 as a brake (left) on the growth-promoting cGMP pathway. The PDE5 blocker sildenafil (blue pills/scissors) releases the PDE5 brake, leading to increased cGMP and MAPK signaling and, ultimately, more aggressive tumor growth (right).

Do all melanoma cells express the growth-promoting cGMP pathway? Apparently not! We detected expression of cGKI and PDE5 in many, but not all, of the human melanoma cell lines tested. Interestingly, primary melanomas isolated from patients expressed substantially higher levels of cGKI⁴ and PDE5¹ than metastases, indicating that CNP and sildenafil act primarily on cells of primary tumors and promote their metastatic potential. However, we only studied the effects of cGKI overexpression and sildenafil on the growth of primary tumors. In the future it will be important to determine whether the cGMP-cGKI pathway and sildenafil do indeed affect melanoma metastasis. Such a role is supported by our finding that the median survival time of patients with high cGKI expression in their tumors is dramatically reduced compared to patients with low cGKI expression.⁴

Do other PDE inhibitors also promote melanoma growth? According to preclinical⁴ and clinical³ data, it is likely that other PDE5 inhibitors like vardenafil (Levitra) or tadalafil (Cialis) have promelanogenic effects similar to those of sildenafil. Moreover, human melanoma cell lines express various levels of other cGMP-hydrolyzing PDEs including PDE1, PDE2, PDE3, PDE9, PDE10, and PDE11.^1,4 Thus, in addition to PDE5, other PDEs and their respective pharmacological inhibitors might modulate the growth-promoting cGMP pathway in melanoma cells.

What are the mechanisms of cGMP-stimulated melanoma growth? We demonstrated that the cGMP pathway interacts with MAPK signaling upstream of MEK and that this crosstalk is required for the growth-promoting effect of cGMP in melanoma cells.⁴ The relevant cGKI substrates and downstream signaling events in melanoma cells have yet to be identified, but there might be parallels to a cGMP-cGKI pathway in vascular smooth muscle cells that also enhances cell growth and survival.⁵ There is growing evidence that the functional outcome of cGMP signaling depends on its compartmentalization as well as on the expression of distinct cGKI isoforms. In the future, a range of melanoma cell lines should be characterized in detail with respect to expression of cGKI isoforms, PDE expression, and their sensitivity to PDE inhibitors and other cGMP-elevating drugs. The existence of functionally distinct cGMP pools, which might be differentially regulated by different PDE blockers, could be analyzed by real-time cGMP imaging with fluorescent biosensors.⁶

How can we integrate increased cGMP signaling into the pathophysiology of melanoma development? The amplitude and duration of cGMP signals are regulated through a dynamic balance between its rate of synthesis by guanylate cyclases and degradation by PDEs. Both processes appear to play a role during melanomagenesis. CNP, acting via GC-B, stimulates cGMP generation in melanoma cells of primary tumors. It is well known that CNP can be secreted by endothelial cells, particularly in the presence of cytokines. Thus, it is likely that the promelanomagenic CNP is released by the inflamed tumor vasculature, providing a new link between inflammation and melanoma. It is conceivable that the level of CNP-induced cGMP in melanoma cells is proportional to the level of tumor vascularization. On the other hand, Houslay recently put forth an elegant model in which a hypoxic melanoma microenvironment generates tumor cells with low PDE5 levels.⁷ The combination of both processes, i.e., increased cGMP generation with increased tumor angiogenesis and decreased cGMP degradation with decreased oxygen supply/vascularization, would provide an auto-regulatory homeostatic mechanism to maintain cGMP concentrations in melanoma cells independent of tumor size and vascularization. Another interesting aspect is that enzymes involved in guanylate metabolism are often altered in human cancers, and recent studies suggest that elevated GTP levels drive melanoma progression through activation of small GTPases such as RAS, RAC, and RHO.⁸ Since GTP is also the substrate of guanylate cyclases, it is tempting to speculate that mutations leading to increased GTP concentrations in melanoma cells may also stimulate the growth-promoting cGMP pathway.

Do PDE5 inhibitors affect other tumors? Preclinical studies have reported both pro- and anticancer effects of cGMP. The variable effects of cGMP on tumor growth are likely due to the fact that different tumor cells express different cGMP generators and effectors. In addition, cGMP may also affect various processes in the tumor microenvironment such as blood flow, angiogenesis, inflammation, and the immune response. Indeed, sildenafil has anti-inflammatory effects and augments endogenous antitumor immunity in several mouse tumor models.⁹ Therefore, it is unlikely that PDE5 inhibitors promote human tumorigenesis in general. It is important to note, however, that recent clinical studies indicate that use of PDE5 inhibitors may be associated not only with melanoma but also with prostate cancer progression.¹⁰

Taken together, our findings in human melanoma cells and mice provide a plausible mechanistic explanation for a causal relationship between sildenafil and melanoma in men. Based on our results we assume that sildenafil, and possibly other PDE5 inhibitors, could first and foremost promote the growth/malignancy of existing melanomas. Thus, possible adverse effects of PDE5 inhibitors should be considered in patients with melanoma, particularly if they use these medications frequently and in high dosages, for instance to treat pulmonary hypertension. Clearly, the available preclinical^1,4 and clinical^2,3,10 data call for more well-controlled prospective clinical studies to confirm a causal relationship between PDE5 inhibitors and melanoma risk.

Disclosure of potential conflict of interest

No potential conflicts of interest were disclosed.

Acknowledgments

I thank M. Paolillo for reading the manuscript and all members of the Feil laboratory for critical discussion.

Funding

This work was supported by the Fund for Science and Deutsche Forschungsgemeinschaft (FE 438/2–4, FOR 2060 projects FE 438/5–1 and FE 438/6–1).