Since its discovery in 1960 by J. W. Black, the non-selective beta-blocker propranolol has been widely used in the treatment of hypertension, tachycardia and other cardiac disorders. More recently, propranolol has been introduced as first-line therapy for infantile hemangiomas, the most common soft-tissue tumors of childhood (Leaute-Labreze et al., 2015).

As propranolol is a lipophilic molecule, its use in infants raised the question of its ability to passively cross the blood-brain barrier (BBB) and directly activate adrenoreceptors on neuronal cells, and to subsequently impact the neurologic development of the child. Even if many reports have documented that propranolol treatment during infancy does not alter further brain development when compared to child and adolescents from the general population examined several years after cessation of treatment (Moyakine et al., 2016, 2017; Gonzalez-Llorente et al., 2017; Mahon et al., 2018), other beta-blockers with hydrophilic properties have been here or there introduced in the treatment of infantile hemangiomas. The most commonly used molecules are nadolol and atenolol, which display hydrophilic properties and are therefore suggested to be unable to cross the BBB (Neil-Dwyer et al., 1981; Dahlof and Dimenas, 1990) and thus potentially decrease the risk to induce deleterious central effects (de Graaf et al., 2013; Pope et al., 2013; Tasani et al., 2017; Alexopoulos et al., 2018). However, these assumptions are based on biophysical characteristics of these molecules, and data regarding long-term safety are clearly lacking.

Alternative mechanisms in which some molecules are able to induce signaling pathways in the brain without crossing the blood-brain barrier have been described. One mechanism, that requires diffusible factors between two partners, implies central neurons and endothelial cells located in vascular wall. Indeed, brain endothelial cells have the capacity to produce second messengers such as nitric oxide (NO) to control the release of neurotransmitters (Knauf et al., 2001). This is especially the case in the mediobasal hypothalamus that includes the median eminence, a circumventricular organ where the BBB is physiologically absent (Rodriguez et al., 2005, 2010; Horsburgh and Massoud, 2013). For example, circulating hormones like estrogen (Prevot et al., 1999) or apelin (Duparc et al., 2011) induce endothelial NO release, with consequences on physiological functions, such as ovulation and glucose homeostasis, respectively. Hydrogen peroxide (H₂O₂) is another way to induce a signaling pathway in this brain region. As previously described for NO signaling, high dose of apelin can modify the release of hypothalamic H₂O₂ that could participate to an over-activation of the sympathetic system leading to the development of a type 2 diabetic state (Drougard et al., 2014). Thus, some molecules can modulate the activity of neuronal cells despite the absence of passage through the BBB and of direct activation of specific receptors on these cells.

The aim of the present study was to investigate whether propranolol, atenolol and nadolol induce the secretion of signaling molecules (i.e., NO and H₂O₂) in the hypothalamus, independently of their ability to directly stimulate adrenoreceptors on neuronal cells.

We have shown that, despite the inability of hydrophilic beta-blockers to cross the BBB and thus to accumulate into the brain, all three beta-blockers tested in this study, either lipophilic or hydrophilic, were able to modulate the release of NO and/or H₂O₂ in the hypothalamus. As these small molecules can passively diffuse in brain tissue, they could have an impact on neuronal activity of central neurons. This could explain some of the potential deleterious effects of these molecules in the brain such as sleep disorders (de Graaf et al., 2013; Pope et al., 2013; Labreze et al., 2015; Leaute-Labreze et al., 2015; Randhawa et al., 2015; Ji et al., 2016) and in periphery such as hypoglycemia (Holland et al., 2010; Poterucha et al., 2015).

Despite its ability to accumulate into the brain, propranolol failed to increase the release of hypothalamic H₂O₂ that is usually associated with oxidative stress (Fisher-Wellman and Neufer, 2012; Angelova and Abramov, 2018). Little is known in the literature to explain the mode of action of propranolol to avoid H₂O₂ release. It has been shown that, in the skin of frog, propranolol is able to decrease water permeability induced by arginine vasotocin (AVT) (Ogushi et al., 2010; Saitoh et al., 2014). In this model, aquaporins are translocated at the membrane in response to AVT. We can speculate that propranolol, by acting on aquaporins translocation, could decrease the negative impact of H₂O₂, which is known to use aquaporins to cross cell membrane (Tamma et al., 2018).

The effect of atenolol on hypothalamic NO/H₂O₂ release is also unexpected. At low dose, atenolol induces NO release from hypothalamic explants, but at higher dose of 250 ng/ml, corresponding to plasma concentration measured after an oral administration at the therapeutic dose, it decreases NO release and increases H₂O₂ secretion. Even if this dual effect seems surprising, our group has previously demonstrated the existence of such physiological phenomenon in another physiological context. Indeed, apelin, an adipokine released by the adipose tissue, stimulates the release of hypothalamic NO at a low dose (Duparc et al., 2011), but inhibits NO release and stimulates H₂O₂ release at a high dose (Duparc et al., 2011; Drougard et al., 2014). In our experimental model, this dual effect can be explained by pharmacological and physiological hypotheses. For instance, the decrease of NO release could be due to its interaction with H₂O₂ to generate hydroxyl radical (Nappi and Vass, 1998).

In addition, we observed that nadolol induces both NO and H₂O₂ secretion either at a low or high dose. The low dose reproduces plasma concentration measured after an oral therapeutic dose of nadolol. This result demonstrates that the ability to cross the BBB is not mandatory to induce signaling in the brain, and thus further activation/inactivation of specific neuronal populations.

Finally, the magnitude of NO/H₂O₂ release in response to the different treatments also needs to be considered. Indeed, while a release in the range of pM to nM induces physiological effects such as signal transduction and neurotransmission (either beneficial or detrimental depending on targeted neurons), a release in the range of nM to μM induces deleterious effects such as oxidative stress and DNA damage leading to cellular dysfunction (Bredt, 1999; Mustafa et al., 2009). In this study we show that, at a concentration mimicking plasma concentration of the three beta-blockers after a therapeutic dose, propranolol induced a release of 80 nM of NO and no change in H₂O₂ production. In contrast, atenolol decreased NO release but increased H2O2 production in the range of μM. Nadolol was responsible of the release of 50 nM of NO and, more importantly, induced a 5 μM release of H₂O₂, which is probably leading to oxidative stress and cellular toxicity. Altogether, these data indicate that, even if they do not cross the BBB and do not accumulate in brain tissues, both atenolol and nadolol are able to induce central toxicity.

Of course, our study is limited by the use of murine ex vivo models which do not reflect the exact physiological conditions observed in vivo. A similar approach (i.e., NO/H₂O₂ real-time measurements) could be performed in vivo in different brain regions in mice in response to an oral load of beta-blockers (Fournel et al., 2017).

To conclude, our study brings new elements to decipher the mode of action (and the potential related side effects) of beta-blockers in the brain. The mode of communication through endothelial cells and production of NO/H₂O₂ that does not require the entry of a beta-blocker molecule into the brain has to be considered when using such molecule.

All authors have designed the experiments and wrote the manuscript. AA had performed the experiments.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.