Occupying a small portion of the vertebrate brain (1%), the PVN is a highly organized effector structure (Swanson, 1995; Benarroch, 2005). This hypothalamic nucleus is located bilaterally around the third ventricle (Badoer, 2001). Morphological studies of the PVN reveal different cell populations within its borders, such that the PVN can be divided into at least three magnocellular (anterior, posterior and medial subnuclei) and five parvocellular (dorsal, lateral, medial, periventricular and anterior subnucleus) compartments (Swanson and Kuypers, 1980; Sawchenko and Swanson, 1982a; Swanson and Sawchenko, 1983; Badoer, 2001; Pyner, 2009). Two functionally separate areas of the PVN, neuroendocrine and autonomic, subserve its potential to regulate BP, making the PVN a major integrative site of cardiovascular function (Swanson and Kuypers, 1980; Swanson and Sawchenko, 1983; Son et al., 2013; Sladek et al., 2015). Blood pressure, which is defined by peripheral vascular resistance and cardiac output (resultant of heart rate and stroke volume), is modified by both neuroendocrine and autonomic premotor PVN in at least three different effector pathways (Figure 1): neurosecretory magnocellular, neurosecretory parvocellular, and pre-autonomic parvocellular neural pathway (Badoer, 2001; Sladek et al., 2015).

Over 30 distinct neurotransmitters and neuromodulators have been identified to be synthesized within the PVN (Swanson and Sawchenko, 1983). Most abundantly expressed are vasopressin (VP) and oxytocin (OT), which are produced by the magnocellular neurons (MCNs) of the PVN (Swanson and Sawchenko, 1980; Sladek et al., 2015). VP is best known for its role in maintaining cardiovascular and body fluid balance, whereas OT takes part in parturition, lactation and accompanying reproductive behaviors (Gimpl and Fahrenholz, 2001; Koshimizu et al., 2006; Stoop, 2014). Additionally, VP and corticotropin-releasing hormone (CRH) expressing parvocellular neurons (PCNs) of the PVN possess a secretory capacity and mediate the central response of the hypothalamic–pituitary–adrenal (HPA) axis to stress (Swanson and Sawchenko, 1980; Sawchenko et al., 1996; Benarroch, 2005; Sladek et al., 2015). Magnocellular and parvocellular neuroendocrine neurons of the PVN initiate a downstream chain of events that dictate changes in BP.

Apart from being locally regulated, the cardiovascular system is subject to central control by numerous relevant brain areas (Dampney, 1994). It is well established that PVN takes an important central place in such control. Numerous studies show that the PVN is implicated in the heightened sympathetic tone observed in hypertension (Allen, 2002; Dampney et al., 2018). The autonomic PVN consists of morphologically and functionally diverse cell populations with a specific topography (Stern, 2001; Dampney et al., 2018). It occupies ventromedial, lateral and dorsal (dorsal cap) subdivision of PVN (Nunn et al., 2011; Sladek et al., 2015). More precisely, these PCNs are pre-autonomic, since they control functionally different sympathetic and parasympathetic centers downstream in the medulla: NTS, dorsal motor nucleus of vagus (DMV) and rostral ventrolateral medulla (RVLM), and the spinal cord: intermediolateral nucleus (IML). Therefore, the “pre-autonomic” premotor PVN is responsible for altering the autonomic output toward the cardiovascular and renal systems (Strack et al., 1989a; Coote et al., 1998; Portillo et al., 1998; Coote, 2005; Dampney et al., 2018). Hence it is not surprising that PVN is commonly referred to as an “autonomic master controller,” a term originally introduced by Loewy (1991). There are at least 3 pathways (Figure 1) through which PVN modulates sympathetic outflow (Pyner, 2009):

The first pathway includes PVN pre-sympathetic neurons that have axonal projections that terminate on somata of motor sympathetic preganglionic neurons (SPNs) in the thoraco-lumbal IML. These neurons are usually named as spinally projecting pre-autonomic neurons (SPANs; Dampney, 1994; Badoer, 2001; Nunn et al., 2011). The IML is the final spot of central integration and origin of preganglionic fibers which regulate the activity of blood vessels, heart, kidneys and adrenal gland. Therefore, this is a particularly important target of the PVN pre-autonomic neurons, displaying its enormous potential to directly alternate neurogenic output to the cardiorenal system (Strack et al., 1989a; Dampney, 1994; Badoer, 2001).

The second pathway involves pre-sympathetic PVN neurons that exert indirect influence on sympathetic activity. These neurons terminate at the level of the motor pressor nucleus of the RVLM, responsible for setting sympathetic tone. From there, second order neurons arise and project to SPNs in the thoracic and lumbar IML and change sympathetic outflow toward the cardiovascular system and the kidneys (Dampney, 1994; Badoer, 2001). In general, PVN neurons projecting to RVLM are in greater number than PVN neurons projecting monosinaptically to IML. Therfore, RVLM is a structure embedding most of the pre-sympathetic neurons, which are projecting to IML to exert major autonomic cardiovascular control (Badoer, 2001). Some of the sympatho-excitatory effects in the PVN-RVLM pathways are glutamate mediated (Yang and Coote, 1998). Additionally, evidence has emerged, that PVN can interfere with this PVN-RVLM pathway through its glutamatergic synapses within medial subnucleus of the NTS (mNTS; Kawabe et al., 2008). Using anterograde and retrograde tracing, Kawabe working group confirmed the presence of such bilateral glutamatergic projections spanning to mNTS. They showed that unilateral PVN stimulation with N-methyl-D-aspartic acid (NMDA) leads to increase in mean arterial pressure and greater splanchic nerve activity, and that this effect is emphasized by bilateral blockade of glutamate ionotropic receptors within the mNTS. The authors suggested that PVN glutamatergic pathways directed toward NTS, stimulate inhibition of RVLM-mediated cardiovascular overactivity. Therefore, PVN stimulation with NMDA evokes both RVLM and NTS neuronal routes, but the effects are opposing (Kawabe et al., 2008). Further, same authors showed that tachycardia induced by NMDA stimulation of the PVN is a result of conjoined inhibition of vagal (via ionotropic glutamate and GABA receptors in the mNTS) and activation of sympathetic outputs (via spinal ionotropic glutamate) toward heart, without involvement of spinal VRs and OTRs (Kawabe et al., 2009).

A third pathway is represented by PVN pre-autonomic neurons that can change sympathetic tone both directly and indirectly. Around 30 % of PVN neurons innervate SPNs in the IML, but send collaterals to the RVLM, thus having a dual regulatory role (Badoer, 2001).

The majority of PVN autonomic regulatory function is conveyed by SPANs (Coote, 2007). Since the relevant portion of SPANs is implicated in cardiovascular control (Badoer, 1996; Badoer, 2001; Pyner, 2009; Nunn et al., 2011), this makes them an attractive target for new drug development.

Although their exact functions have yet to be elucidated, SPANs have been suggested to regulate blood volume (Lovick et al., 1993; Pyner and Coote, 2000), circadian variations in BP (Cui et al., 2001), stress induced cardiovascular responses (Jansen et al., 1995a) and much more (Nunn et al., 2011). For these reasons SPANs are referred to as “central command neurons” (Jansen et al., 1995a). Discovering their primary physiological function is further complicated by the fact that SPANs express a lot of different neuroactive substances/neurotransmitters. Indeed, it appears that the majority of these neurons synthesize more than one neurotransmitter (Nunn et al., 2011). The largest portion of SPANs (up to 40 %) is positive for VP and OT (usually co-expressed), as well as dynorphin (Hallbeck and Blomqvist, 1999; Xi et al., 1999; Hallbeck, 2000; Hallbeck et al., 2001). Others possess met-endorphin (up to 20%) and dopamine, met-enkephalin (up to 10 %), leu-enkephalin, somatostatin, ANG II and ANP (Sawchenko and Swanson, 1982a; Cechetto and Saper, 1988; Strack et al., 1989b; Jansen et al., 1995b).

Functional in vivo studies show that spinal levels of VP and OT increase with PVN stimulation (Pittman et al., 1984; Malpas and Coote, 1994). The presence of V1aR and OTR has been confirmed in the gray matter of the spinal cord (Desaulles et al., 1995; Sermasi et al., 1998; Pyner, 2009). Intrathecal pretreatment with V1aR antagonist in the lower thoracic region prevents the increase in renal sympathetic nerve activity and mean arterial pressure normally triggered by the stimulation of the PVN (Malpas and Coote, 1994). It cannot be excluded that some of the OT effects are conveyed via spinal V1aRs as well (Sermasi and Coote, 1994). Similar effects are also observed with OTR selective antagonists which abolish the effects of increased heart rate following the PVN stimulation (Yang et al., 2009). These experimental data highlight VP and OT as PVN SPAN transmitters. VP and OT induce cardio-acceleratory and pressor effects in lower and upper thoracic spinal cord respectively (Pyner, 2009). The majority of vasopressinergic projections implicated in cardiovascular control originate in the lateral and ventral medial subdivision of parvocellular PVN and some portion is found in dorsal parvocellular part (Riphagen and Pittman, 1989; Hallbeck and Blomqvist, 1999). Oxytocinergic SPANs are located in the lateral parvocellular region and dorsal cap (Sawchenko and Swanson, 1982a; Pyner, 2009).

The effects of dopamine as a putative SPAN neurotransmitter are controversial (Strack et al., 1989b; Nunn et al., 2011). Some studies on rats suggest its excitatory influence on SPNs, while others report an inhibitory activity (Gladwell et al., 1999; Gladwell and Coote, 1999a,b; Yang et al., 2002).

Pharmacological experiments with a glutamate antagonist, following chemical stimulation of PVN, imply glutamate as an additional excitatory neurotransmitter of SPANs (Yang et al., 2002).

Although SPANs are positive for enkephalins and ANG II, their influence on sympathetic outflow has not yet been recorded (Sawchenko and Swanson, 1982a; Cechetto and Saper, 1988; Jansen et al., 1995b; Hallbeck and Blomqvist, 1999; Hallbeck et al., 2001). Despite its abundant expression in the nervous system, the presence of inhibitory γ-aminobutyric acid (GABA) as a SPAN output neurotransmitter has not been confirmed (Macdonald and Olsen, 1994; Watkins et al., 2009).

The peripheral sympathetic nervous system controlling the cardiovascular system has rhythmic activity that creates distinct patterns of sympathetic nerve discharge (SND) in response to physiological demands and pathophysiological conditions. Using frequency analysis of SND, components of SND were identified including fast cardiac and respiratory rhythms, and slow vasomotor rhythm. It is generally believed that the brain is the source of SND. Thus, which parts of the brain are involved, and how they generate peripheral SND rhythms is still unanswered. Two main theories have been postulated, both of which raised considerable criticism. A theory of a central oscillator/pacemaker suggests that the RVLM is the main structure responsible for generating peripheral sympathetic discharge patterns via axonal projections to preganglionic neurons in IML. Recordings of intracellular neuronal activity from medullary slices, uncovered ramp like depolarization following each action potential, leading to subsequent action potential suggestive of pacemaker activity (Li and Guyenet, 1996). This observation was seriously challenged by the in vivo study in rats by Lipski et al. (1996) who observed that RVLM neurons fire irregularly, at much higher discharge rates than the ones in SND, with no evidence of gradual depolarization between individual action potentials (Lipski et al., 1996). They suggested that the regular pattern of firing of RVLM neurons seen in medullary slices, was produced by deafferentation. The work of Lipski and colleagues imposes the network hypothesis, where the activity of pre-sympathetic neurons depends on their antecedent excitatory inputs opposed by tonic inhibitory inputs (setting the level of their excitability). For detailed review refer to Malpas (1998). Geber and Barman postulated a theory of a network of brainstem neurons, whose combined action creates inherent rhythmicity entrained by the baroreflex. These include sympatho-excitatory and sympatho-inhibitory neurons distributed over a wide portion of the lower brainstem that do not have necessarily spontaneous activity (Barman and Gebber, 2000).

In addition to RVLM neurons, hypothalamic PVN SPANs provide monosynaptic inputs to preganglionic neurons in IML. Malpas and Coote (1994) were first to provide evidence in anesthetized rats that chemical stimulation of the PVN by microinjections of homocysteic acid increases the amplitude of renal SND, by recruiting more active fibers, and that this can be blocked by intrathecal injection of V1R antagonist. At the same time the frequency of renal SND, which is modulated by periodic baroreflex inputs (Ninomiya et al., 1990; Malpas and Ninomiya, 1992), remained unaffected. The first experimental evidence that PVN SPANs can incite BPV at the same stimulation frequency was demonstrated in rats by Stauss and associates (Stauss and Kregel, 1996; Stauss et al., 1997). PVN electrically stimulated at frequencies between 0.1 Hz - 0.5 Hz were found to create the same frequencies in the SND pattern and generate BPV in the low frequency (LF) band, which is abolished by α-adrenergic blockade (Japundzic et al., 1990). Stimulation frequencies above 0.5 Hz did not induce BP variations as the blood vessels behaved like cut off filters to high frequencies. Studies on isolated rat vascular smooth muscle cells showed that transmission of fast SND rhythms to BPV is limited by the sluggish, metabotropic α-adrenoceptor signaling, and not by an intrinsic inability of the cells to contract and relax at higher rates (Julien et al., 2001). Thus, SND frequencies higher than 0.5 Hz in rats induce vasoconstriction, increase peripheral resistance and the mean value of BP. It follows that the faster components of BPV, the cardiac component and the respiratory component, are non-neural, created by the perturbations of the circulation induced by the contracting heart and inspiratory movements (Japundzic et al., 1990; Japundzic-Zigon, 1998). The non-neural origin also stands for the slowest, and dominant component of BPV, the very low frequency (VLF) component, which was found to be created by inherent myogenic activity of mesenteric and renal vasculature (VanBavel et al., 1991; Janssen et al., 1995). Although non-neuronal in nature, all the components of BPV can be modulated by the activity of the nervous system and neurohormones (Japundžić-Žigon et al., 2020).

We have investigated the neurochemical contribution of the PVN to short-term BPV, and found that both VP and OT modulate short-term BPV. Using pharmacological and genetic tools in conscious rats we have shown that VP modulates BPV in a complex manner: peripherally as a hormone and centrally as neurotransmitter/modulator (Japundzic-Zigon, 2001, 2013; Japundzić-Zigon et al., 2004, 2018, 2020; Milutinović et al., 2006a,b; Stojicić et al., 2008; Milutinović-Smiljanić et al., 2013; Lozić et al., 2016; Savić et al., 2020). Using spectral analysis of BPV, we found that peripheral administration of non-peptide and selective V1aR or V2R antagonists to conscious normotensive rats under baseline physiological conditions increases BPV, suggesting a buffering role for VP in the VLF domain (Japundzic-Zigon, 2001). We postulated that the decrease of VLF-BPV by VP could be mediated either by the enhancement of the baroreflex sensitivity which normally opposes VLF-BPV, or by the modulation of vasomotion in mesenteric and renal vascular beds (Janssen et al., 1995). In SHR, the buffering capacity of VP on BPV under baseline condition is not preserved (Japundzić-Zigon et al., 2004). This could be due to pathological remodeling of the vasculature in SHR (Head, 1991; Vågnes et al., 2000) making it more sensitive to vasoconstrictors, including VP. Another possibility, even more likely is that impaired baroreflex in SHRs (Dampney, 2017) reduces its capacity to buffer VLF-BPV (Dampney, 2017). However, during hemorrhage, when VP is released in excess in blood, in support of circulation, it acts similarly in normotensive and hypertensive rat strains, and prevents the respiratory related high frequency (HF-BPV) increase, possibly as a consequence of V1aR-mediated vasoconstriction which prevents the unloading of thoracic vessels underlying HF-BPV increase.

Intracerebroventricular injection of selective V1aR, V1bR and V2R antagonists to conscious normotensive rats uncovered that VP acts also centrally to buffer VLF-BPV under baseline physiological conditions by the stimulation of V1aRs possibly in AP, accessible from both sides of the blood brain barrier. However, when VP release is stimulated by stress (Stojicić et al., 2006, 2008; Milutinović et al., 2006b), or by drugs (Milutinović et al., 2006a) and when VP is injected centrally (Milutinović et al., 2006b), an increase of the sympathetically mediated LF-BPV and of the respiration mediated HF-BPV was observed (Milutinović et al., 2006a,b). These effects of VP could involve central V1aR found in abundance in the RVLM where integration of the sympathetic outflow to vasculature occurs; and in the pre-Bötzinger area, where the breathing pattern is set. These central effects of VP could be beneficial and contribute to the lifesaving effect of VP in hemorrhagic, septic and cardiogenic shock as a results of increased tissue oxygenation and additional sympathetic activation, acting synergistically with powerful V1aR mediated peripheral vasoconstriction (Levy et al., 2018). VP can also modulate respiration indirectly (Stojicić et al., 2008) by an anxiogenic action characterized by hyperventilation that occurs possibly by the stimulation of V1bR in bed nucleus stria terminalis (Griebel et al., 2002).

Paraventricular nucleus is a recognized key integrative site of the behavioral, autonomic and endocrine response to stress, expressing V1aRs and V1bRs on somata and dendrites of MCNs and surrounding glia. Stress has been shown to induce VP and OT release in the PVN too (Nishioka et al., 1998; Landgraf and Neumann, 2004), and we have shown that VP and OT act locally, in an autocrine and paracrine manner, to modulate the neuro-cardiogenic stress response. Using adenoviral gene transfer technology we have increased the gene expression and the number of V1aRs in the PVN of Wistar rats. The V1aR rat phenotype had decreased sensitivity of the baroreflex under baseline physiological conditions which was further decreased by stress along with a marked increase of the sympathetically mediated LF-BPV and LF heart rate variability (LF-HRV). These effects could be abolished by intranuclear application of V1aR antagonist. This suggests that V1aR in the PVN can increase the sympathetic outflow to the periphery and modulate BPV and HRV during stress (Lozić et al., 2014; Japundžić-Žigon et al., 2020). In clinical practice, the increase in LF-HRV has been found to predict the occurrence of life threatening arrhythmias in susceptible populations (Huikuri and Stein, 2013). In contrast to V1aR, OTR over-expression in the PVN of Wistar rats had no effect on BPV under baseline conditions but this rat phenotype exhibited reduced baroreflex desensitization by stress and reduced LF-BP increase suggesting that OTR over-expressing rat phenotype is resilient to stress (Lozić et al., 2014; Japundžić-Žigon et al., 2020). In this context, it is important to stress that the modulation of the baroreceptor desensitization during stress by VP seems to be complex and involves more than one central structure and type of VRs (Milutinović-Smiljanić et al., 2013).

A number of clinical studies in hypertensive patients unequivocally show that enhanced BPV increases the risk for developing cardiovascular complications (Mancia and Grassi, 2000). Thus, BPV, and especially sympathetically derived LF-BPV, emerged as an independent predictor of stroke, coronary artery disease, heart and renal failure, as well as all-cause mortality (Mancia et al., 1994; Mancia and Parati, 2003; Messerli et al., 2019; Parati et al., 2020). Our group investigated expression of PVN VP and VRs in the genesis of hypertension. Borderline hypertensive rats (BHR) have a genetic predisposition for hypertension and will develop it when exposed to environmental challenges. Under baseline physiological conditions BHRs have increased expression of VP and V1bR in the PVN and consequently increased plasma VP concentrations, as a constitutive trait (Savić et al., 2020). Spectral markers of sympathetic activity toward blood vessels, LF-BP, and the heart, LF/HF-HR, are comparable to normotensive rats under baseline physiological conditions, suggesting that increased expression of V1bR and VP in BHRs is confined to magnocellular (endocrine) portion of the PVN affecting plasma VP only. However, when exposed to repeated stress and prolonged isotonic saline load, BHRs exhibited LF-BPV increase depicting sympathetic overload, and overt hypertension. In these rats no changes in VP and VR gene transcription in the PVN was noted. Moreover, systemic VP release was decreased, refuting involvement of VP in stress-induced hypertension (Savić et al., 2020).

It is well established that PVN has a paramount role in cardiovascular regulation and contributes to the severity of cardiovascular diseases. Both neuroendocrine and autonomic PVN, have a dynamic part in adjusting the circulation to physiological demands and in the modulation of short-term BPV. Thus, elucidating the tightly intertwined mechanisms underlying complexity of the PVN network in health and disease may open up new therapeutic venues.

BS, DM, and NJ-Ž: outlining paper draft, writing and refining the manuscript, and critical reading of the manuscript. All of the authors have read and approved the manuscript.

This work was supported by Serbian Ministry of Education, Science and Technological Development (no. 200110, BS, NJ-Ž), British Heart Foundation (RG/11/28714, DM; FS/12/5/29339, DM), and BBSRC (BB/J005452/1, DM).

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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.