Activation of the vasopressinergic neurons and release of AVP into the circulation occurs in response to osmotic and non-osmotic stimuli. The magno- and parvocellular vasopressinergic cells of the hypothalamus are distinctly activated by various stimuli, with magnocellular cells being mostly activated by increase in sodium ion concentration and osmolality, hypovolemia, hypotonia, and hypoxia, while the parvocellular cells are more sensitive to various stressors, such as pain, injury, and psychological stress (Aguilera et al., 2008; Ueta et al., 2011; Bankir et al., 2017; Szczepanska-Sadowska et al., 2017). Next, we discuss main stimuli for AVP release and how they affect the respiratory system.

Vasopressin is released into the circulation in response to increase in plasma osmolality and sodium ion concentration in directly proportional manner (Verney, 1947; Quillen and Cowley, 1983; Thornton et al., 1986; Baylis, 1987; Verbalis, 2007). In young healthy men, plasma AVP concentration starts to increase at plasma osmolality of 285mOsm/kg, and continues to rise till values of osmolality exceed 310mOsm/kg (Baylis and Robertson, 1980; Baylis, 1987). The changes in osmolality are detected in the organum vasculosum of the lamina terminalis (OVLT), in the subfornical organ (SFO), and in the magnocellular neurons of the PVN and the SON (Anderson et al., 2000, 2001; Verbalis, 2007; Bankir et al., 2017; Szczepanska-Sadowska et al., 2017). The changes of extracellular fluid osmolality are mostly affected by changes in sodium ion concentration; however, other osmotically active substances (e.g., mannitol) that affect cell volume also trigger AVP release (Verbalis, 2007). On the other hand, osmotically active solutes freely passing through the cell membrane (e.g., urea), have much weaker effect on cell volume and AVP release (Verbalis, 2007). In addition, changes in activity of AVP neurons and subsequent release of AVP in the posterior pituitary lobe are regulated by availability of water and sensation of thirst and may precede actual changes in the extracellular fluid osmolality (Bankir et al., 2017).

Several studies indicate that plasma osmolality and concentration of sodium ions may contribute to the control of respiration. Specifically, hyperosmolality was shown to inhibit respiration and to reduce increase in pulmonary ventilation evoked by thermoregulatory panting or adjustments to acid-base balance disturbances in animals (Baker and Dawson, 1985; Kasserra et al., 1991; Kasserra and Jones, 1993) and humans (Senay, 1969; Moen et al., 2014). However, it should be noted that experiments in the in situ preparations revealed direct excitatory effects of hyperosmolality on the carotid sinus nerve and the phrenic nerve activity (Trzebski et al., 1978; Kasserra et al., 1991; O’Connor and Jennings, 2001; da Silva et al., 2019).

In contrast to hyperosmolality, acute hypoosmolality has been shown to stimulate breathing in conscious dogs (Anderson et al., 1990) and transiently in rats (O’Connor and Jennings, 2001). In women and men, hypoosmolality induced by ingestion of water also increases ventilation, although, this may be in part compensatory response to metabolic acidosis induced by ingestion of tap water (Moen et al., 2014).

Earlier experiments in vivo on cats have shown that perfusion of the carotid bodies with hypoosmotic solutions increases activity of the carotid sinus nerve (Gallego and Belmonte, 1979) that is associated with increase in breathing. This stimulatory effect of hypoosmolality, at least in part, may be attributed to activation of calcium currents and depolarization of chemoreceptors in the carotid body directly induced by hypoosmotic stimulus, as shown in a rat (Molnár et al., 2003). Another potential mechanism may depend on vasoconstriction of vessels supplying the carotid body, leading to decrease in the carotid body blood flow and activation of the chemosensitive glomus cells (Brognara et al., 2021). It has been shown that hypoosmolality causes constriction of the vascular smooth muscles in various vascular beds (Lang et al., 1995; Aoki et al., 2014), and this mechanism was postulated to be involved in the activation of carotid bodies and increased activity of the carotid sinus nerve (Gallego and Belmonte, 1979).

Vasopressinergic neurons can be activated and AVP secreted into the circulation in response to numerous non-osmotic stimuli. The most critical ones are hypovolemia and hypotension, physical exercise, hypoglycemia, hypoxia, and Ang II. Of note, these stimuli also affect the pulmonary ventilation and/or activate the carotid body and arterial chemoreflex.

It has been shown that V1aRs are predominantly expressed in the lung (Hirasawa et al., 1994; Tahara et al., 1998) and in the pulmonary arteries (Enomoto et al., 2014). Contrary to arteries in the systemic circulation, AVP does not cause vasoconstriction of the pulmonary arteries, but rather it causes vasodilation, especially under conditions of hypoxic vasoconstriction of pulmonary vessels in animals (Walker et al., 1989; Wallace et al., 1989; Enomoto et al., 2014; Sugawara et al., 2019) and humans (Jeon et al., 2006; Currigan et al., 2014). Thus, AVP does not increase pulmonary vascular resistance and the right ventricle’s afterload, but it only selectively increases the vascular resistance in systemic circulation (Walker et al., 1989; Jeon et al., 2006).

Retrograde transneuronal labeling indicates that the airway-related vagal preganglionic neurons receive innervation from vasopressinergic neurons of the PVN (Kc et al., 2006, 2010). Furthermore, AVP acting on V1aRs depolarizes and increases firing rate of these preganglionic neurons (Yan et al., 2017), which is suggestive of the broncho-constrictive and secretory effect of AVP. Nonetheless, available evidence points to limited effects of AVP on bronchoconstriction in laboratory animals (Bhoola et al., 1962; Zheng et al., 2017) and humans (Knox et al., 1989).

Recently, we showed presence of immunostaining for both C-terminal and N-terminal fragments of the V1aR in the chemosensitive glomus cells of the carotid bodies collected from the normotensive Sprague-Dawley rats (Żera et al., 2018). Furthermore, expression of G protein q/11 and phosphokinase C, key intracellular components of the V1aR signaling, has been detected in the chemosensitive glomus cells with the single-cell transcriptomics in mouse (Zhou et al., 2016). These findings indicate that circulating AVP may affect the glomus cells expressing V1aRs and presumably modulate the activity of the carotid body.

Circumventricular organs (CVOs) include the organum vasculosum laminae terminalis (OVLT) and the SFO that are located close to the third cerebral ventricle, and the area postrema (AP), which is situated in the dorsal surface of the medulla oblongata. They express numerous receptors, lack the blood-brain barrier and are accessible for hormones and mediators circulating in the bloodstream (Ufnal and Skrzypecki, 2014). All CVOs express receptors for AVP, predominantly V1aRs (Raggenbass et al., 1989; Jurzak and Schmid, 1998; McKinley et al., 1999; Tribollet et al., 1999; Kc and Dick, 2010; Hindmarch et al., 2011). Upon binding to V1aRs in the CVOs of the third ventricle, AVP exerts sympathoexcitatory and pressor effects (Ufnal and Skrzypecki, 2014; Szczepanska-Sadowska et al., 2017; Japundžić-Žigon et al., 2020). In contrast, stimulation of V1aRs in the AP sensitizes the arterial baroreflex and results in hypotension (Hasser and Bishop, 1990; Japundžić-Žigon et al., 2020), which indicates opposite effects of AVP depending on the place of its action. A limited number of studies indicates that administration of AVP into the cerebral ventricles or directly into the AP decreases respiratory rate and phrenic nerve activity (Zerbe and Feuerstein, 1985; Yang et al., 2006).

In rats, V1aRs are expressed on the sympathetic neurons of the rostral ventrolateral medulla (RVLM) and the respiratory neurons located in the rostral ventral respiratory column (rVRC; Kc and Dick, 2010) and pre-Bötzinger complex (Kc et al., 2002a). Retrograde labeling studies show that in addition to these nuclei, the phrenic nuclei receive vasopressinergic innervation from the PVN (Kc et al., 2002a,b). The expression of V1aRs in these brainstem structures is augmented by exposure of the animals to hypoxia (Kc and Dick, 2010; Kc et al., 2010). Furthermore, a significant subpopulation of the vasopressinergic PVN neurons are activated by hypercapnia (Kc et al., 2002b). Importantly, binding of AVP to V1 receptors was found in neurons across the nucleus of the solitary tract (NTS), including its caudal part (Raggenbass et al., 1989) that receives sensory input from the arterial chemoreceptors (Lipski et al., 1976, 1977; Finley and Katz, 1992; Zera et al., 2019).

Intravenous infusions of AVP transiently decrease pulmonary ventilation and phrenic nerve activity in conscious dogs (Ohtake and Jennings, 1993), anesthetized and awake rats (Louwerse and Marshall, 1993; Walker and Jennings, 1998; Żera et al., 2018; Brackley and Toney, 2021) and fetal lambs (Bessho et al., 1997). Recently published pilot study in patients with autosomal dominant polycystic kidney disease showed that upon initiation of treatment with tolvaptan, a selective V2R antagonist, plasma copeptin level increased 6-fold and this was associated with a modest but significant increase in arterial carbon-dioxide and plasma acidity (Heida et al., 2021), suggestive of ventilatory inhibition by increased AVP levels in the bloodstream and enhanced stimulation of V1 receptors, which were not blocked by tolvaptan.

The inhibitory effects of AVP on the respiratory activity depend on V1aRs, as blockade of these receptors with selective antagonists completely prevents changes in pulmonary ventilation induced by systemic administration of AVP in anesthetized rats (Louwerse and Marshall, 1993; Żera et al., 2018; Brackley and Toney, 2021). Furthermore, administration of V1aR antagonists may result in the increase in pulmonary ventilation in awake dogs (Walker and Jennings, 1994). However, it should be noted that insignificant effect of V1aR blockade on pulmonary ventilation was seen in conscious rats (Walker and Jennings, 1998). Furthermore, intravenous administration of V1aR peptidergic antagonist, that does not cross the blood-brain barrier, has insignificant effects on hypoxia-induced increase in the pulmonary ventilation in conscious dogs (Overgaard et al., 1996), suggesting lack of involvement of V1aRs located in the peripheral organs or CVOs in the increase in ventilation evoked by activation of the arterial chemoreflex in this species. However, systemic administration of V1aR antagonist unmasks stimulatory effects of Ang II on the respiration in dogs (Walker and Jennings, 1995). Interestingly, oxytocin, which is structurally related to AVP and acts as a weak agonist of the V1aRs (Szczepanska-Sadowska et al., 2017), is capable of reversing the opioid-induced respiratory depression and this effect is fully unmasked by blockade of V1aRs (Brackley and Toney, 2021).

One of the plausible explanations of the inhibitory action of systemic administration of AVP on ventilation is a baroreflex-mediated inhibition of the medullary respiratory neurons (Richter and Seller, 1975; Brunner et al., 1982; Baekey et al., 2010; McMullan and Pilowsky, 2010) due to the AVP-mediated rise in arterial blood pressure. However, as mentioned above the arterial baroreflex-mediated inhibition of the pulmonary ventilation in the conscious rat is vaguely pronounced (Walker and Jennings, 1996). Thus, another mechanism is presumably at play.

Based on the experimental work, Jennings (1994) suggested that angiotensin II and AVP present in the bloodstream may affect ventilation via interaction with the CVOs. Thus, one of such mechanisms may involve stimulation of the V1aRs in the CVOs, specifically in the AP. In addition to the AP neurons responding to both AVP and the rise in blood pressure, indicating their pressure-sensitivity, there are at least two pools of neurons in the AP that respond to circulating AVP independently from the changes in arterial blood pressure, indicating their AVP-sensitivity. One pool increases and another one decreases the firing rate in response to intravenous AVP (Smith et al., 1994). Both types of responses are dependent on vasopressin V1 receptors, as their inhibition abolishes AVP-induced changes in the discharge frequency (Smith et al., 1994). Furthermore, in the medial regions of the NTS there are neurons that respond with increased or decreased firing in response to AVP microinjections into the AP (Cai et al., 1994), indicating that the circulating neurohormone may affect activity of the NTS involved in the baro- and chemoreflexes via the neuronal pathway including the AP. In fact, application of AVP into the AP significantly reduces the phrenic nerve activity in mechanically ventilated, urethane anesthetized rats, and the inhibitory effect is prevented by local administration of V1aR antagonist into the AP or by pharmacological inhibition of the NTS (Yang et al., 2006). This is further supported by observations that AVP administered into the cerebral ventricles, thus allowing for interaction of the neurohormone with the CVOs, decreases respiratory rate in the anesthetized rat (Zerbe and Feuerstein, 1985). However, it should be noted that intracerebroventricular infusion of AVP in conscious macaque monkey had no effect on ventilatory parameters (Lee et al., 1985). This is contrary to the studies with electrical and pharmacological stimulation of the AP neurons in rabbits, which results in increased phrenic nerve activity (Srinivasan et al., 1993; Bongianni et al., 1994). It should be noted that electrical stimulation of the AP may trigger both excitation or inhibition of the NTS neurons in rodents (Chrobok et al., 2021). Similarly, electric stimulation of the SFO increases ventilation in rats (Ferguson et al., 1989). Although, the V1aRs are expressed in the SFO (Ostrowski et al., 1994), thus far it has not been determined how their stimulation affects the respiratory system. Available evidence and anatomical proximity to the respiratory neurons of the brainstem point to the AP as a putative CVO involved in mediating the respiratory effects of circulating AVP. However, given the limited number of studies that evaluated respiratory responses to AVP directly administered into the CVO, the role of CVOs in neural control of respiration by AVP awaits further corroboration from experiments specifically targeting AVP and the AP and/or the SFO, especially under conditions associated with activation of vasopressinergic system and in awake animals.

Circulating AVP may also affect control of the respiratory system by binding to its receptors within the carotid body. We showed that in normotensive rats, AVP administered locally into the carotid body via the internal carotid artery causes a modest increase in the pulmonary ventilation without significant changes in the blood pressure (Żera et al., 2018). Further studies are needed to determine whether these effects are directly dependent on activation of the chemosensitive glomus cells that express V1aRs (Żera et al., 2018) or rather on AVP-mediated decrease in the carotid body blood flow that may sensitize the chemoreceptors (Przybylski, 1981; Brognara et al., 2021).

Retrograde labeling and functional studies show neural pathways between the PVN and the NTS, the RVLM, and the presympathetic neurons of the spinal cord (Yang and Coote, 1998; Yang et al., 2002; Ferguson et al., 2008; Affleck et al., 2012). The PVN is involved in the peripheral chemoreflex, as the inhibition or disinhibition of the PVN neurons result in the respective attenuation or augmentation of the peripheral chemoreflex-evoked sympathetic and respiratory responses (Reddy et al., 2005). Furthermore, retrograde labeling studies combined with immunostaining for V1aRs show that the PVN vasopressinergic fibers terminate on neurons in the RVLM, the rVRC, the pre-Bötzinger complex, the NTS, and the phrenic nuclei (Kc et al., 2002a, 2010; Jackson et al., 2005; Figure 1). In a series of experiments, Kc et al. (2002b) showed that hypercapnia activates vasopressinergic neurons in the PVN and hypoxia upregulates V1aRs in the RVLM, the ventral respiratory group and in the phrenic nuclei in the spinal cord (Kc et al., 2010). They also showed that in anesthetized mechanically ventilated and vagotomized rats, disinhibition of the PVN leads to increase in respiratory activity, as estimated by means of electromyography of the diaphragm and genioglossal muscle and that this increase can be prevented by a pre-treatment of the RVLM and the rVRC with selective V1aR antagonist (Kc et al., 2010). Along with the upregulation of V1aRs, the respiratory responses were augmented by chronic intermittent hypoxia (Kc et al., 2010). The increases in respiratory activity evoked by disinhibition of the PVN were accompanied by pressor response and dependent on V1aRs (Kc and Dick, 2010, Kc et al., 2010). Together, these studies indicate that activation of the vasopressinergic PVN projections to the RVLM/rVRC increases the respiratory activity via V1aRs.

In addition, microinjections of AVP into the pre-Bötzinger complex or RVLM/rVRC also increased the diaphragm’s muscle activity in a V1aR-dependent manner (Kc et al., 2002a, 2010). The increases in respiratory activity evoked by local application of AVP was accompanied by pressor response (Kc et al., 2002a). Furthermore, microinjection of the V1aR antagonist into the RVLM/rVRC resulted in a decrease in respiratory activity in rats exposed to chronic intermittent hypoxia (Kc et al., 2010), suggestive of tonic vasopressinergic input to the respiratory neurons. Together, these studies indicate that AVP and V1aRs in the RVLM and rVRC are involved in stimulation of the respiratory activity.

Contrary effects were reported in a series of experiments, in which AVP was microinjected into various sites of the ventrolateral medulla (VLM) caudal from the RVLM determined as a pressor region of the VLM in urethane anesthetized and vagotomized rats that were paralyzed and mechanically ventilated. In these experiments, microinjections of AVP applied into the VLM and the rostral ventral respiratory group (rVRG) caudal from the RVLM resulted in apnoea with subsequent inhibition of the phrenic nerve discharges (Chuang et al., 2003, 2005). Although, the inhibitory effect of AVP on the respiratory activity was consistent, the pressor response was dependent on the site of injection. Applications of AVP into the lateral part of the VLM decreased the phrenic nerve activity and concomitantly elevated arterial blood pressure, whereas AVP application into the medial part of the VLM attenuated only the phrenic nerve activity, without significant effect on the arterial blood pressure (Chuang et al., 2003, 2005; Cheng et al., 2004). Microinjections of AVP into the VLM and the rVRG were also shown to inhibit activity of the hypoglossal nerve (Chuang et al., 2005). Of note, these inhibitory effects of AVP on phrenic nerve activity were diminished by hypercapnia (Chuang et al., 2003; Cheng et al., 2004) and were completely abolished by pre-treatment with a selective V1aR antagonist (Chuang et al., 2003, 2005; Cheng et al., 2004). Administration of the V1aR antagonist under resting conditions had insignificant effect on blood pressure and phrenic nerve activity in these experiments (Chuang et al., 2003, 2005; Cheng et al., 2004), suggesting lack of significant impact of vasopressinergic efferents innervating these structures on ventilation at rest. Together, these findings suggest inhibitory action of AVP on the respiratory neurons of the rVRG located caudally from the RVLM.

These two series of reports provide apparently conflicting results that may be caused by the several differences in the experimental paradigms, including use of neuromuscular blockade and atropine. In addition, mechanical ventilation with continuous positive end-expiratory pressure potently increases plasma concentration of AVP (Annat et al., 1983; Venus et al., 1985) and this activation of AVP release depends on the hydration status of the ventilated animal (Venus et al., 1985). Thus, various levels of baseline activity of vasopressinergic system could have been present across the above studies. It should be also noted that acute experiments under anesthesia investigating the autonomic responses related to the function of the brainstem and the hypothalamus may yield conflicting results to physiological responses observed in conscious animals (Kannan et al., 1989). Nonetheless, it seems that in the more rostral part of the VLM/rVRC AVP promotes respiratory activity, whereas in the more caudal area, it inhibits respiration.

The carotid body-evoked hyperglycemic response has been recognized as a contributing factor in the pathophysiology of metabolic syndrome (Conde et al., 2017) and blockade of V1aRs in the NTS was shown to attenuate increase in the plasma concentration of glucose induced by activation of the peripheral chemoreflex (Montero et al., 2006). Moreover, inhibition of the V1aRs in the NTS substantially reduces pressor and tachycardic effects of electrical stimulation of the PVN (Pittman and Franklin, 1985). Thus, it is probable that AVP in the NTS also modulates respiratory reflexes, especially those originating from the peripheral chemoreceptors. This assumption is further substantiated by evidence indicating that almost half of the NTS neurons respond with excitation to local application of AVP in the coronal sections of the rat’s brainstem (Raggenbass et al., 1989). Furthermore, in anesthetized cat, iontophoretic applications of AVP into the NTS revealed excitatory effects of the neuropeptide on the inspiratory neurons of the NTS (Henry and Sessle, 1989). However, microinjections of AVP into a discrete region of the NTS rostral to the calamus scriptorius also produced inhibition of the phrenic nerve activity in rats (Yang et al., 2006). Further investigations are required to elucidate the significance of AVP action in the NTS for regulation of respiration.

The role of endogenous AVP and vasopressinergic neurons in the regulation of the respiratory system is also indirectly supported by spectral analysis of the blood pressure variability, the heart rate variability, and selective pharmacological inhibition of AVP receptors in the central nervous system in awake rats, which suggests that vasopressin and V1aR/V1bRs are involved in enhancing ventilation and respiratory-induced blood pressure oscillations (Japundzic-Zigon, 2001; Milutinović et al., 2006). Furthermore, experiments in the Brattleboro rats lacking AVP indicate that although under resting conditions their breathing is normal, under challenging conditions of septic shock their respiratory adaptation, i.e., increase in respiration, is absent (Brackett et al., 1983), which suggests the important role of the neuropeptide in respiratory adaptation to disturbed homeostasis. This is further supported by known respiratory problems, including sleep apnoea, decreased ventilation and spells in patients with Wolfram syndrome characterized by diabetes insipidus and a lack of AVP (Thompson et al., 1989; Licis et al., 2019).