A non-canonical chemical feedback self-limits nitric oxide-cyclic GMP signaling in health and disease

Endothelial nitric oxide (NO) stimulates the heme protein, soluble guanylyl cyclase (sGC) to form vasoprotective cyclic GMP (cGMP). In different disease states such as pulmonary hypertension, NO-cGMP signaling is pharmacologically augmented, yet the pathomechanisms leading to its dysregulation are incompletely understood. Here we show in pulmonary artery endothelial cells that endogenous NO or NO donor compounds acutely stimulate sGC activity, but chronically down-regulate both sGC protein and cGMP formation. Surprisingly, this endogenous feedback mechanism was independent of canonical cGMP signaling via cGMP-dependent protein kinase. It did not involve thiol-dependent modulation, a process relevant for sGC maturation, either. Rather tonic NO exposure led to inactivation and degradation of NO-sGC and without affecting NO-insensitive apo-sGC levels. Apo-sGC could be re-activated pharmacologically by the heme mimetic class of so-called sGC activators. Importantly, this non-canonical feedback was also observed in vivo. Specifically, it was induced by pathological high levels of NO in acute respiratory distress syndrome in which a similar self-limiting redox shift from NO-sensitive sGC to NO-insensitive apo-sGC occurred. Thus, our data establish a bimodal mechanism by which NO acutely stimulates sGC and chronically decreases sGC levels as part of a physiological and pathological self-limiting feedback. Of therapeutic importance in disease, our findings i) caution against any chronic use of classical NO donor drugs and ii) suggest that high NO-induced apo-sGC can be functionally fully recovered by sGC activator drugs to re-establish cGMP formation. Significance Statement Dysfunctional nitric oxide (NO) signaling via the cyclic GMP (cGMP) forming heme-protein soluble guanylate cyclase (sGC) is a key cardiopulmonary disease mechanism. However, in particular during chronic use, NO donor drugs display limited therapeutic benefit. Here we identify a previously unrecognized non-canonical chemical feedback mechanism, which selflimits cGMP formation in response to NO donor drugs and endogenous NO, both in health and disease. Whilst NO acutely stimulates sGC, we find that exposure of sGC to either chronic NO or pathological NO overproduction reduces sGC and generates heme-free apo-sGC which is insensitive to NO but sensitive to heme-mimetic sGC activators. Importantly, this chemical feedback explains the limited applicability of NO-donor drugs for chronic treatment and explain their mechanism-based indication in vascular disease conditions.


Abstract (240 words)
Endothelial nitric oxide (NO) stimulates the heme protein, soluble guanylyl cyclase (sGC) to form vasoprotective cyclic GMP (cGMP). In different disease states such as pulmonary hypertension, NO-cGMP signaling is pharmacologically augmented, yet the pathomechanisms leading to its dysregulation are incompletely understood. Here we show in pulmonary artery endothelial cells that endogenous NO or NO donor compounds acutely stimulate sGC activity, but chronically down-regulate both sGC protein and cGMP formation. Surprisingly, this endogenous feedback mechanism was independent of canonical cGMP signaling via cGMPdependent protein kinase. It did not involve thiol-dependent modulation, a process relevant for sGC maturation, either. Rather tonic NO exposure led to inactivation and degradation of NO-sGC and without affecting NO-insensitive apo-sGC levels. Apo-sGC could be re-activated pharmacologically by the heme mimetic class of so-called sGC activators. Importantly, this noncanonical feedback was also observed in vivo. Specifically, it was induced by pathological high levels of NO in acute respiratory distress syndrome in which a similar self-limiting redox shift from NO-sensitive sGC to NO-insensitive apo-sGC occurred. Thus, our data establish a bimodal mechanism by which NO acutely stimulates sGC and chronically decreases sGC levels as part of a physiological and pathological self-limiting feedback. Of therapeutic importance in disease, our findings i) caution against any chronic use of classical NO donor drugs and ii) suggest that high NO-induced apo-sGC can be functionally fully recovered by sGC activator drugs to reestablish cGMP formation.

\body Introduction
The NO-cGMP signaling pathway plays an important role in cardiopulmonary homeostasis (1,2). The best-defined receptor and mediator of NO's actions is soluble guanylate cyclase (sGC), a heterodimeric heme protein. During enzyme maturation, NO facilitates heme incorporation into sGC (3,4). In its Fe(II)heme-containing state, sGC binds NO and is thereby activated to convert guanosine triphosphate (GTP) to the second messenger, cGMP (5), which exerts its cardiopulmonary effects via cGMP-dependent protein kinase (6). This results in potent vasodilatory, anti-proliferative and anti-thrombotic effects (7). In disease, heme loss and appearance of an apo-sGC has been described (8) (9).
Activation of sGC by NO can cause acute and, importantly, reversible desensitization involving possibly protein S-nitrosylation (10). Chronic exposure to NO donor drugs has been suggested to negatively affect sGC more long-term and in a not fully reversible manner (11)(12)(13), but it is unclear how this effect is mediated and whether it also pertains to endogenously formed NO.
Here, we examine this important knowledge gap in porcine pulmonary artery endothelial cells (PPAECs) as this excellently relates to the clinical relevance of NO and cGMP modulating drugs in pulmonary hypertension (14,15). We compare the effects of chronic exposure to exogenous NO donor drugs and endogenous NO generated by endothelial NO synthase both on sGC protein and cGMP formation. In addition, we investigate i) whether any of the underlying mechanisms involves canonical cGMP signalling, thiol modulation, or NO-insensitive apo-sGC and ii) whether this would be relevant in disease. For the latter, we use a well-established high-NO model of porcine acute respiratory disease syndrome (ARDS) related also to pulmonary hypertension (16).
Our findings provide important new understandings of NO-cGMP signaling.
Pharmacologically, we identify i) previously not recognized risks of chronic use of NO donor drugs and ii) how chronically high-NO disease conditions lead to cGMP deficiency that can be recovered in a mechanism-based manner through heme mimetic, apo-sGC activator drugs.

Endogenous NO down-regulates vascular sGC protein and activity in vivo and in vitro.
We first tested whether endogenous NO instead of only pharmacological NO donor compounds downregulate sGC upon chronic exposure. Previously, protein levels were nonconsistently investigated or the antibodies used were of unclear specificity (11,17). Moreover, the functional consequences of cGK-I on cGMP levels were investigated only in some cases (18,19) and in these cases rather due to upregulated cGMP metabolism rather than an effect on sGC activity (18,20,21). We therefore, initially studied this using in vivo models, i.e. i) endothelial NO synthase wildtype or knock-out (eNOS -/-) mice, which are characterized by physiological or decreased NO levels, respectively or ii) in the porcine lung disease model of lavage-induced acute respiratory distress symptom (ARDS) which is characterized by NO overproduction (22).
In eNOS -/mice, sGCa1 and sGCb1 levels were up-regulated (Fig. 1A). This up-regulation of sGC protein subunits was associated with increased NO-stimulated sGC-activity (Fig. 1B). In ARDS, sGCa1 and sGCb1 protein levels were modulated albeit, this time modulation occurred in the opposite direction (Fig. 1C) and resulted in decreased sGC-mediated cGMP production (Fig. 1D).
Collectively, these data suggest that in vivo low NO led to compensatory up-regulation of sGC subunits and pathological high levels of endogenous NO effectively down-regulate sGC.
Next, we investigated whether PPAECs could be a suitable in vitro model to mechanistically characterize our in vivo observations. In line with the observations in eNOS -/mice, 72h incubation of PPAECs in presence of the NO synthase (NOS) inhibitor N G -nitro L-arginine methyl ester (L-NAME), increased sGCb1 levels whilst sGCa1 levels remained unchanged ( Fig.  2A). This up-regulation of sGCb1 protein was associated with increased sGC-activity (Fig. 2B).
To test the converse, we increased PPAECs NO levels by incubating the cells with supraphysiological amounts of NO using the long-acting NO donor compound, DETA/NO (100 µM).
Indeed, incubation with DETA/NO decreased sGCa1 and sGCb1 protein levels ( Fig 2C) and decreased DEA/NO-induced cGMP production (Fig. 2D). Thus, in PPAECs low levels of NO increase sGC expression and activity and high levels of NO down-regulate sGC.

NO.
Having established the effect of chronic/tonic NO on sGC protein and activity, we aimed to clarify the underlying mechanisms. Indeed, cGK-I, has been shown to reduce sGC mRNA levels (18). Cell passaging can cause downregulation of cGMP-dependent protein kinase-I (cGK-I), a critical component in the NO-cGMP pathway (23)(24)(25)(26). Hence, we therefore restricted our studies to low passage number cells and ensured fully functional cGK-I signaling by validating that the NO donor, DETA-NO, the cGK-I activator, 8-Br-cGMP, or the NO-independent Fe(II)sGC stimulator were able to reduce cGK-I expression (Fig. 3A), an expressional regulation known to be cGMP/cGK-I-dependent (27).
sGK-I-mediated, cGMP-dependent cGK-I auto-regulation has been shown (28). Thus, having established functional cGK-I signaling in PPAECs, we studied whether a similar mechanism affects sGC expression by exposing PPAECs for 72h with different concentrations of YC-1 or 8-Br-cGMP. However, none of these compounds affected sGC protein levels like NO (cf. to Fig. 2).
In fact, we observed an up-regulation of sGC ( We subsequently aimed to validate our in vitro findings in vivo using cGK-I knock-out (cGK-I -/-) mice (29). Indeed, cGK-I -/mice displayed normal sGC expression levels (Fig. 4A) and unchanged sGC activity (Fig. 4B) as compared to wildtype controls. The apparent lower sGC activity level in cGK-I knock-out mice did not reach significance, but would be consistent with the up-regulation observed with cGK-I stimulators observed in vitro. In conclusion, our in vivo and in vitro data suggested that down-regulation of sGC expression by NO is both cGMP-and cGK-I-independent and thus non-canonical.
To assess thiol-dependence of the sGC down-regulation we incubated PPAECs with DETA-NO (100µM for 72h) in absence or presence of the membrane-permeable thiol-reducing agent, N-acetyl-L-cysteine (NAC; 1mM for 72h). Presence of NAC did not affect sGCa1 and sGCb1 protein expression (Fig. 5A), nor NO-stimulated sGC activity (Fig. 5B). These data suggested that it is unlikely that a thiol dependent desensitization-like mechanism (10,33,34) is involved in chronic NO-induced reduction in sGC expression/activity.
To examine whether chronic NO exposure causes formation of NO-insensitive apo-sGC we took advantage of one of the apo-sGC activator drugs (i.e. BAY 58-2667) which specifically bind to the empty heme binding pocket of apo-sGC and mimic heme sterically, with respect to its charge distribution, and regarding heme's effects on propionic acid substituents (35). Again, to simulate chronic NO exposure we incubated PPAECs with DETA-NO (100 µM for 1 to 72h) after which we assayed BAY 58-2667-stimulated (apo-)sGC activity. Indeed, NO incubation increased BAY 58-2667-induced apo-sGC activity within 1h and this increased stability remained stable for at least 72h post-DETA-NO treatment (Fig. 6A). In parallel, DETA/NO incubation reduced NO-stimulated sGC activity within 1 h and this reduction was stable over time (Fig. 6A).
To validate this important mechanistic finding in vivo, we used the porcine ARDS model. Consistent with our in vitro findings, NO-stimulated sGC activity in ARDS animals was strongly reduced (Fig. 1D) while apo-sGC activity increased to a smaller extent (Fig. 6B). This was prominently evidenced by a dramatic shift in the cGMP forming capacity from NO-sensitive sGC towards apo-sGC (Fig. 6C). These data suggested that NO signaling by a self-limiting downregulation of sGC is caused omission of NO-sensitive sGC so that NO-insensitive, but BAY 58-2667-sensitive apo-sGC remains. The higher protein levels in the presence of BAY 58-2667 can be explained by the fact that apo-sGC is prone to proteolytic degradation and stabilized by apo-sGC activator compounds (36,37).
In conclusion, our data suggest that both in vitro and in vivo, both under physiological conditions and in disease NO appears to self-limit its ability to induce cyclic GMP formation via a chemical redox feedback which causes inactivation of sGC and subsequent generation, and degradation of NO-insensitive apo-sGC. Thus, our findings explain i) why cGMP synthesis is diminished in the context of chroming treatment with NO-donor drugs and ii) why apo-sGC activator compounds such as BAY 58-2667 should be considered as an alternative to NO-donors (4,8).

Discussion
Here, we establish and characterize a previously unrecognized redox mechanism exerting an irreversible negative feedback within NO-cGMP signaling both under physiological and, more so, under pathological high NO conditions. NO not only facilitates sGC's maturation but also induces a small tonic degradation pressure generating also low levels of apo-sGC.
With respect to the underlying mechanism, we initially considered the two canonical mechanisms that are known to mediate or modulate NO-cGMP signaling, i.e. activation of cGK-I following an increase in cGMP, and short-term, reversible desensitization of sGC by thiol modification (4,10). Surprisingly, both could be excluded. Instead, our finding indeed suggest a novel mechanism of sGC regulation reminiscent of an earlier observation where a longer term (20 h) exposure to an NO donor reduces sGC activity irreversibly and in a manner that could not be recovered by thiol treatment (13). The availability of sGC activator compounds now allowed us to fully elucidate the mechanism as related to a combination of sGC loss and apo-sGC formation. Our present study thereby fills an important knowledge gap in the pharmacology of sGC and establishes a third mechanism in sGC regulation by chronic NO. In health and in disease, this leads to a shift in the sGC redox ratio from NO-sensitive sGC to NOinsensitive apo-sGC.  (40), where we could now clarify the mechanism of the underlying dysfunctional cGMP signaling. We postulate that this will also explain why cGMP-signaling is impaired in other forms of high-NO conditions, such as stroke (41) and neurotrauma (42).
Our findings are also of direct therapeutic importance as a pathological sGC/apo-sGC ratio can be treated with sGC activator compounds such as BAY58-2667 (36) thereby reinstalling cGMP synthesis and cGK-I signaling (32,43). However, with respect to the long-established class of NO donor drugs a cautionary not needs to be added. Not only do they cause tolerance, but, as we now find, also irreversible downregulation of sGC and apo-sGC formation; this may explain the superiority of novel, NO-independent sGC modulation drugs in pulmonary hypertension, such as sGC stimulators (15).

Materials and Methods
Chemicals. Polyclonal antibodies specific for sGCa1 and sGCb1 have been described elsewhere (30). Actin monoclonal antibody (Oncogene Research Products, Boston, USA); collagenase type CLS II (Biochrom, Berlin, Germany); 8-bromo-cGMP (BIOLOG, Bremen, Germany); L-NAME, DETA/NO and DEA/NO (Alexis, Lausen, Switzerland); IBMX and GTP (Enzo LifeSciences, Lörrach, Germany); YC-1 was a kind gift from Dr. Schönafinger, Aventis Pharma (Frankfurt, Germany). BAY 58-2667 was synthesized as described (44) . Preparation of pulmonary arteries from a porcine ARDS model. Pulmonary arteries were removed immediately after death from an experimental porcine model of acute respiratory disease symptom (ARDS), as previously described (33). Pulmonary arteries were snap-frozen in liquid nitrogen and stored at minus 80°C or otherwise processed immediately to tissue powder and subsequently suspended in homogenization-buffer and homogenized in an Ultra Turrax at 4°C. These samples were then used further for protein determination, protein immune blots and sGC activity assays.