Transient receptor potential vanilloid type 4 is expressed in vasopressinergic neurons within the magnocellular subdivision of the rat paraventricular nucleus of the hypothalamus

Abstract Changes in plasma osmolality can drive changes in the output from brain centres known to control cardiovascular homeostasis, such as the paraventricular nucleus of the hypothalamus (PVN). Within the PVN hypotonicity reduces the firing rate of parvocellular neurons, a neuronal pool known to be involved in modulating sympathetic vasomotor tone. Also present in the PVN is the transient receptor potential vanilloid type 4 (TRPV4) ion channel. Activation of TRPV4 within the PVN mimics the reduction in firing rate of the parvocellular neurons but it is unknown if these neurons express the channel. We used neuronal tracing and immunohistochemistry to investigate which neurons expressed the TRPV4 ion channel protein and its relationship with neurons known to play a role in plasma volume regulation. Spinally projecting preautonomic neurons within the PVN were labelled after spinal cord injection of FluoroGold (FG). This was followed by immunolabelling with anti‐TRPV4 antibody in combination with either anti‐oxytocin (OXT) or anti‐vasopressin (AVP). The TRPV4 ion channel was expressed on 63% of the vasopressinergic magnocellular neurosecretory cells found predominantly within the posterior magnocellular division of the PVN. Oxytocinergic neurons and FG labelled preautonomic neurons were present in the same location, but were distinct from the TRPV4/vasopressin expressing neurons. Vasopressinergic neurons within the supraoptic nucleus (SON) were also found to express TRPV4 and the fibres extending between the SON and PVN. In conclusion within the PVN, TRPV4 is well placed to respond to changes in osmolality by regulating vasopressin secretion, which in turn influences sympathetic output via preautonomic neurons.

different cell groups with distinct neuro-hormonal functions (Swanson & Sawchenko, 1983). Two morphologically distinct cell groups have been described: magnocellular neurosecretory cells (MNCs) that synthesise and release the peptide hormones AVP and oxytocin (OXT) and project exclusively to the posterior pituitary; and parvocellular neurons, that are further subdivided into two groups: one which releases corticortropinreleasing factor to evoke the release of adrenocorticotropic hormone from the anterior pituitary (Antoni, 1993) and another group, the preautonomic neurons that influences autonomic function via projections to the brain stem and spinal cord . The spinally projecting preautonomic neurons have also been categorised as intermediate and named mediocellular (Kiss, Martos, & Palkovits, 1991). The preautonomic neurons influence blood pressure, heart rate and sympathetic nerve activity, including renal sympathetic nerve activity.
The transient receptor potential vanilloid type 4 channel (TRPV4) is a non-selective cation channel transducing physical stress, for example, osmotic cell swelling or mechanical stress into intracellular Ca 2+ −dependent signalling (Sharif-Naeini, Ciura, Zhang, & Bourque, 2008). More recently, it has been demonstrated that TRPV4 is activated by increased cell volume irrespective of the molecular mechanism underlying cell swelling and thus the channel is suggested to function as a volume-sensor, rather than (or as well as) an osmosensor (Toft-Bertelsen, Krizaj, & MacAulay, 2017).

The TRPV4 channel may be involved in systemic osmoregulation
and there is some evidence to support a physiological role for TRPV4 in the hypothalamic osmosensing nuclei (Liedtke & Friedman, 2003;Carreño, Ji, & Cunningham, 2009). The TRPV4 channel is expressed in the PVN and SON where it is co-localised with AVP containing cells (Carreño, Ji, & Cunningham, 2009). Functionally, TRPV4 and calcium activated potassium (K Ca ) ion channels have been shown to couple as osmosensors in the PVN in mouse brain slices and rat isolated PVN neurons (Feetham, Nunn, Lewis, Dart, & Barrett-Jolley, 2015b). Again in mice, intracerebroventricular administration of hypotonic artificial cerebrospinal fluid decreases blood pressure but not heart rate and inhibition of the TRPV4 ion channel attenuated this effect (Feetham, Nunn, & Barrett-Jolley, 2015a). While these studies demonstrate a functional role for TRPV4 in osmosensing within the PVN, they do not establish which neurons express the channel, leaving open the question of the neuronal mechanism underlying these observations. Here we have used retrograde labelling of spinally projecting preautonomic neurons in combination with immunohistochemistry for TRPV4, AVP and OXT to determine where TRPV4 protein is expressed and its relationship to cell groups involved in osmosensing and cardiovascular homeostasis within the PVN.

| Ethical Approval
All experiments were approved by the Local Ethics Committee of Durham University and performed in accordance with UK Animals (Scientific Procedures) Act, 1986 and the European Commission Directive 86/609/EEC (European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes). All surgical procedures were carried out on anaesthetised animals that minimised suffering with the minimum number of animals used. Animal were killed with an overdose of sodium pentobarbital (60 mg/kg) at the termination of the experiment.

| Injection of retrograde tracers
Six male wistar rats were anaesthetised intraperitoneally with medetomidine 0.25 mL/100 g and ketamine 0.06 mL/100 g prior to spinal cord injection of the retrograde tracer FluoroGold (FG; Fluorochrome -Denver, Colorado, USA LLC). The FG (2% in 0.9% saline) was pressure injected into the left intermediolateral region of the spinal cord at the T2 level (Watkins, Cork, & Pyner, 2009). Following injection, analgesia was administered (0.01 mL/100 g buprenorphine) and the animals recovered for 7-10 days with ad libitum food and water.

| Perfusion-Fixation
After the recovery period, animals were terminally anaesthetised and perfused with heparinised saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4). Brains and spinal cord were removed, post fixed overnight at 4 C and then transferred to 30% sucrose-phosphate buffer (4 C) until sectioned.

| Immunohistochemistry
Immunohistochemistry was carried out on free floating sections cut on a freezing microtome at 40 μm. Transverse sections of PVN were collected at the levels containing centres engaged in cardiovascular control (Swanson & Sawchenko, 1983; and longitudinal sections of spinal cord (100 μm) were used to confirm the location of the injection site within the intermediolateral cell column (Swanson & Sawchenko, 1983;. Nonspecific binding sites were blocked with 10% normal goat serum (NGS; Abcam Cambridge CB4 0FL, UK, Ab7481)-0.1% Triton-X-100 (TX) in PB for 45 minutes, rinsed in PB (1 x 10 mins) then incubated in rabbit anti-TRPV4 (1:400 in 1% NGS-0.1% TX in PB; Abcam 94,868 lot GR276084, RRIDAB_10675981) overnight at 4 C. Four animals underwent double labelling for anti-TRPV4

| Confocal Microscopy
Sections were examined using a Zeiss 880 Laser Scanning Confocal Microscope. Images were captured using Zen 2.1 SP2 (black; version 13.0.2.518). Frame mode acquisition was utilised to capture FluoroGold (excitation 405 nm, emission 530-600 nm), Alexafluor 488 (excitation 488, nm emission 494-600 nm) and Alexaflour 594 (excitation 594 nm, emission 604-735 nm). Overview images were captured using x20 objective (NA 0.8) in tile scan mode to generate the large field of view required and z stacks as required. Regions of interest were subsequently imaged with either x40 or x63 oil objectives (NA 1.3 and 1.4 respectively). Raw images were processed using Zen (blue edition) software and final images were imported into Adobe Photoshop (CS4 extended v. 11.02) to create annotated figures.

| Cell Counts
Cell counts were generated using a cell counter plugin in the Javabased image processing program ImageJ (https://imagej.nih.gov/ij/, 1997-2016.). FluoroGold labelled neurons were counted in consecutive sections throughout the rostrocaudal extent of the PVN, from approximately Bregma −1.40 to −2.12, ipsilateral to the spinal cord injection site. Abercrombie's correction for double counting errors was applied to these counts (Abercrombie, 1946). In four animals alternate sections were labelled with anti-TRPV4 and AVP (every other section receiving the TRPV4 and OXT combination). Therefore cell counts of TRPV4 and AVP labelled populations were obtained from alternate sections. As the effective size of these sections was 80 μm, no correction was made for double counting errors.

| Antibody Specificity
These are all commercial antibodies subject to routine quality assurance (Table 1). Where positive results were obtained the pattern of reactivity was characteristic of that particular antibody with distinct cell populations consistently labelled by that antibody on repeat assays. There was an absence of labelling with secondary antibodies alone. For the anti-TRPV4 antibody a further antigen preadsorption control was included. Prior incubation of the antibody with the immunising peptide (Abcam 230,486 1 mg/mL, 1:1 with antibody overnight at 4 C) abolished labelling (Fig 1bc). Although preadsorption controls for the antibodies to OXT and AVP were not undertaken, both these antibodies have been used in previous, published studies to identify MNC's in rat PVN (Nedungadi & Cunningham, 2014;Reuss, Brauksiepe, Disque-Kaiser, & Olivier, 2017).

| Spinally projecting preautonomic neurons
The injection site was identified and confirmed as being in the left intermediolateral cell column in all those animals whose tissues were used for immunohistochemistry ( Figure 1a

| TRPV4 ion channel expressing neurons
The

| Relationship between TRPV4 and spinally projecting neurons
Rarely were neurons expressing TRPV4 identified as spinally projecting preautonomic neurons apart from in one case found in the caudal region of the PVN and this neuron was also vasopressinergic

| Relationship between TRPV4 and vasopressinergic neurons
Neurons immunoreactive for vasopressin were predominantly con-

| Relationship between spinally projecting-and vasopressinergic neurons and spinally projecting-and oxytocinergic neurons
Only occasionally were the spinally projecting neurons found to be vasopressinergic ( Figure 5: Y asterix). A similar pattern for OXT was observed, however while only a small number of spinally projecting preautonomic neurons were immunoreactive for OXT it was more than for AVP (Figure 3: X asterix). The spinally projectingoxytocinergic neurons were predominantly located within the caudal  Spinally projecting preautonomic neurons in the parvocellular region of the PVN did not express TRPV4 or contain AVP and a topgraphical segregation was evident between the spinally projecting and the TRPV4-AVP neurons. Topgraphical segregation between AVP and PVN-RVLM projecting neurons has also been reported (Son et al., 2013). Only one preautonomic neuron was found to be immunoreactive for both TRPV4 and AVP. In situ hybridisation suggests that 40% of spinally projecting neurons contain mRNA for AVP. In colchicine treated rats at least 35% of the PVN-spinally projecting neurons con- V1a receptor (Gilbey, Coote, Fleetwood-Walker, & Peterson, 1982;Backman & Henry, 1984;Ma & Dun, 1985;Malpas & Coote, 1994;Sermasi, Howl, Wheatly, & Coote, 1998). The fact that rarely do PVN parvocellular neurons contain AVP seems at odds with the functional evidence and it may reflect that within the parvocellular neurons AVP is stored as neurophysin (Swanson, 1977;White, Krause, & McKelvy, 1986).
Putative dendrites from the TRPV4-AVP neurons were seen to project into and through the preautonomic neurons and closley appose them. Again a similar pattern has been reported for PVN-RVLM projecting neurons (Son et al., 2013). As all TRPV4 neurons are vasopressinergic and these neurons are a separate population to the parvocellular neurons irrespective of autonomic target, then we can suggest that the PVN-RVLM would not express TRPV4. Indeed a recent study provides evidence for another TRP family member  (Hallbeck, Larhammar, & Blomqvist, 2001) while for those parvocellular neurons projecting to the stellate ganglion about 10% have been shown to be oxytocinergic (Jansen, Wessendorf, & Loewy, 1995). Functionally, OXT has been implicated in autonomic regulation with a direct action on sympathetic preganglionic neurons (Gilbey, Coote, Fleetwood-Walker, & Peterson, 1982;Yasphal, Gauthier, & Henry, 1987;Sermas & Coote, 1994;Deusaules, Reiter, & Feltz, 1995;Yang, Wheatley & Coote, 2002). While the primary roles of AVP and OXT are very different, oxytocinergic MNCs can also be activated by increasing osmolality but our evidence would indicate the TRPV4 ion channel is not part of the mechanism (Leng et al., 2001;Oliveria et al., 2004).

| TRPV4-parvocellular interaction
The paraventricular nucleus of the hypothalamus has been shown to be critical to sensing and responding to changes in plasma osmolality (Bourque, 2008). Disturbances in osmolality and the evoked cellular response involve TRPV4-activation coupled to the low-conductance calcium-activated potassium (SK) channel. An in vitro study using mouse brain slices and rat isolated PVN neurons demonstrated that anatomically and morphologically defined parvocellular neurons responded to osmolality (Feetham, Nunn, Lewis, Dart, & Barrett-Jolley, 2015b). Superfusion of the brain slices with hypotonic artificial cerebrospinal fluid was found to reduce action current frequency and these effects were mediated by coupling of TRPV4/SK channels. Similarly, an in vivo study investigated whether hypotonic TRPV4 driven neuronal inhibition modulated cardiovascular parameters. In mice, intracerebroventricular administration of hypotonic solutions decreased mean blood pressure but not heart rate and inhibition of the TRPV4 channels abolished these effects (Feetham, Nunn & Barrett-Jolley, 2015a). These studies support a central TRPV4 channel as important for sensing osmolality and the authors proposed the effects of its activation to be mediated by the channel expressed on spinally projecting preautonomic neurons (Feetham, Nunn, Lewis, Dart, & Barrett-Jolley, 2015b Release of AVP from MNCs is closely related to electrical activity of these cells (Leng, Brown, & Russell, 1999). Magnocellular neurosecretory cells appear to positively correlate their rate of action potential discharge with extracellular fluid osmolality (Bourque, 1998).
Under normal body fluid osmolality, the firing rate of the MNCs (~1-3 Hz) mediates basal AVP secretion, whereas hypotonicity and hypertonicity decrease and increase respectively, firing frequency and AVP release (Bourque, 1998).
Recently crosstalk between AVP MNCs and PVN-RVLM projecting preautonomic neurons has been proposed (Son et al., 2013). Activity dependent dendritic release of AVP from neurosecretory neurons has FIGURE 5 Transient receptor potential vanilloid 4 is expressed exclusively on AVP neurons within the PVN. Panel c4 from Figure 4 with high magnification insets to show that TRPV4 is present on the majority of vasopressinergic (AVP) cells (inset X) and spinally projecting preautonomic neurons are separate from these cell groups (inset Y). Expanded inset X: The majority of AVP neurons express TRPV4 (AVP/TRPV4*), however 37% of AVP neurons do not express TRPV4 (**). Expanded inset Y: A single spinally projecting preautonomic neuron immunoreactive for AVP and TRPV4 (*). This was the only occasion on which a triple labelled cell was seen. Otherwise spinally projecting preautonomic neuron (FG, blue) were always distinct from the AVP/TRPV4 (orange) and AVP (green) cells. Fine varicose fibres labelled with TRPV4 radiate laterally from the AVP/TRPV4 cells (arrows). On occasion these fibres are closely apposed to the spinally projecting preautonomic neurons (arrowheads). Abbreviation: 3 V-3rd ventricle [Color figure can be viewed at wileyonlinelibrary.com] been shown to stimulate PVN-RVLM preautonomic neurons demonstrating a mechanism for interpopulation crosstalk (Son et al., 2013). The released AVP is proposed to act as a diffusible signal between populations of neurons within the PVN. A central osmotic challenge administered via intracarotid infusion of increasing concentrations of sodium chloride results in concentration dependent increases in renal sympathetic nerve activity (Chen & Toney, 2001). This effect was shown to be due to intranuclear release of AVP from MNCs because blockade of the V1a receptor within the PVN blunted the increase in renal sympathetic activity suggesting the AVP was contributing to the sympathoexcitation (Son et al., 2013). Thus central osmotic challenge is able to modulate sympathetic output; and separate studies have shown that this can be attenuated centrally either by TRPV4 inhibition (Feetham, Nunn & Barrett-Jolley, 2015a) or blockade of V1a receptors on spinally projecting preautonomic neurons (Son et al., 2013). Our demonstration of TRPV4 on AVP MNCs suggests these effects may be mediated via release of AVP from these neurons onto preautonomic sympathetic neurons with which they are in close proximity.

| Conclusion
Our study provides an anatomical understanding of how changes in osmolality may affect sympathetic output from spinally projecting neurons within the PVN of the hypothalamus. Osmotic sensing by the TRPV4 ion channel expressed on AVP MNCs may lead to dendritic release of AVP and its subsequent diffusion onto preautonomic net- works. The precise mechanism for sensing and signalling of osmotic disturbances and thus blood plasma volume has important implications in heart failure as recent evidence would suggest AVP modulation of sympathoexcitibility is impaired in disease. A reduction in the expression of SK (small conductance K ca channels) in hypothalamic MNCs in heart failure rats has been shown to contribute to the hyperexcitibility of those neurons (Ferreira-Neto, Biancardi & Stern, 2017).
An increase in hypothalamic MNC excitability could lead to increased AVP release and the well documented sympathoexcitation observed in heart failure animals (Abboud, 2010). Functional studies in combination with anatomical analyses detailing the location of osmosensitive proteins on cell groups within cardiovascular control centres and their interconnections, is gradually revealing the mechanisms underpinning cardiovascular homeostasis.

CONFLICT OF INTEREST STATEMENT
The authors declare they have no conflict of interest. In memory of Professor John H Coote .

AUTHOR CONTRIBUTIONS
The authors take responsibility for the integrity of the date and accuracy. Study concept and design FCS and SP. Acquisition of data FCS.
Analysis and Interpretation of data FCS and SP. Writing of manuscript FCS and SP.