Heart failure‐induced reprogramming of salt‐evoked neurovascular coupling in the rat hypothalamus Authors: Ranjan K. Roy, Elba Campos Lira, Manuel Bita Ongolo, Jessica A. Filosa, Javier E. Stern Published: 2026 ABSTRACT AbstractNeurovascular coupling (NVC) is essential for matching cerebral blood flow to neuronal activity. While NVC has been extensively studied in the cerebral cortex, far less is known about its regulation in deep brain regions such as the hypothalamus, particularly under pathological conditions. Heart failure (HF) is associated with impaired cortical NVC and cognitive dysfunction, but its impact on hypothalamic NVC remains unclear. Here, we investigated salt‐evoked NVC in the hypothalamic supraoptic nucleus (SON) of rats with HF. In vivo two‐photon imaging combined with real‐time tissue oxygen measurements was used to assess parenchymal arteriole diameter and local PO2 in the SON during a systemic hypertonic saline challenge in a rat model of HF. Pharmacological interventions and biosensor approaches were employed to dissect the contributions of key vasoactive signalling pathways. Under baseline conditions, HF rats exhibited reduced SON PO2 compared with sham animals. Unexpectedly, salt loading in HF rats elicited a vasodilatory NVC response accompanied by a progressive increase in SON PO2, restoring oxygen levels toward those observed in sham rats. This polarity shift was mediated by adenosine signalling through A2A receptors and depended on intact vasopressin (AVP) signalling. Notably, blockade of A2A receptors unmasked an underlying AVP‐dependent vasoconstriction, indicating the coexistence of competing vasoactive mechanisms during salt challenge. These findings demonstrate a state‐dependent reorganization of hypothalamic NVC in HF, in which adenosine‐dependent vasodilatation predominates over AVP‐mediated vasoconstriction during salt loading. This adaptive shift highlights the dynamic regulation of neurovascular signalling in the hypothalamus and may have broader implications for neurohumoral control in cardiovascular disease. Key points Neurovascular coupling (NVC) is well characterized in the cerebral cortex but remains poorly understood in the hypothalamic nuclei regulating homeostatic and neurohumoral functions.Heart failure (HF) is associated with a basal hypoperfused and hypoxic state of the hypothalamic supraoptic nucleus (SON), indicating impaired local neurovascular regulation at rest.During systemic salt loading, NVC in the SON of HF rats undergoes a polarity switch, with vasodilatation and improved tissue oxygenation replacing the vasoconstrictive response in sham animals.This state‐dependent reorganization involves adenosine signalling via A2A receptors and is functionally dependent on intact vasopressin/V1A receptor signalling, indicating that opposing vasoactive pathways jointly shape the net vascular response to salt loading in HF.These findings highlight dynamic, region‐specific regulation of hypothalamic neurovascular signalling in cardiovascular disease, with implications for neurohumoral control during homeostatic challenge. Introduction Neurovascular coupling (NVC) is a well‐established phenomenon in the brain that involves a tight coupling between neuronal activity and increased cerebral blood flow (CBF) (Iadecola, 2017; Urban et al., 2017). This activity‐dependent increase in CBF (also known as functional hyperaemia) occurs rapidly (<2 s) and in a spatially restricted manner (Li et al., 2009; Silva et al., 2000), ensuring the optimal delivery of glucose and oxygen to brain areas with increased metabolic demand. NVC is generally thought to be a uniform process across brain regions and is typically mediated by byproducts of synaptic activity (Meyer‐Baese et al., 2025). However, we recently demonstrated a unique form of NVC in the hypothalamic supraoptic nucleus (SON), in which increased neuronal activity of neurosecretory vasopressin (AVP) neurons, triggered by a systemic salt challenge, induced an activity‐dependent vasoconstriction (termed inverse NVC or iNVC) that led to local tissue hypoxia and further increase in AVP neuronal excitability (R. K. Roy et al., 2021). Furthermore, we demonstrated that the iNVC was mediated by a local somatodendritic release of AVP, and that this positive feedback loop played an important role in the ability of these neurons to homeostatically respond to the peripheral interoceptive challenge (R. K. Roy et al., 2021). A compromised NVC function is typically observed in several neurodegenerative disorders, such as Alzheimer's disease and other forms of dementia (Wareham et al., 2022), contributing to cognitive deficits in these diseases. Notably, NVC is also compromised in cardiovascular diseases, including heart failure (HF) (Aires et al., 2020; Houben et al., 2003), hypertension (Lefferts et al., 2018; Youwakim et al., 2023) and stroke (Lin et al., 2011). A recent study in a mouse model of HF showed an impaired NVC response in the somatosensory cortex in response to whisker stimulation that contributed to cognitive impairment and dementia (Hoyer‐Kimura et al., 2024). Moreover, a compromised NVC in the cerebral cortex was demonstrated in HF patients with both reduced ejection fraction (EF) and preserved EF (Aires et al., 2020; Lambrichts et al., 2025). Still, whether altered NVC responses in other brain regions relevant to the pathology of HF also occur in this disease, such as the hypothalamus, remains unknown. A hallmark of HF is neurohumoral activation, characterized by increased sympathoexcitatory output to the heart, kidneys and vasculature, along with a systemic release of neurohormones, including AVP. Importantly, neurohumoral activation is directly associated with overall morbidity, mortality and survival rates for HF (Cohn et al., 1984). Several hypothalamic mechanisms have been shown to contribute to increased neurohumoral drive in HF, including increased AVP intrinsic neuronal excitability (Patel et al., 2000; Reis et al., 2016), changes in the excitatory/inhibitory balance (Potapenko et al., 2011), increased expression and release of AVP (Patel et al., 2000; Price et al., 2004), and blunted nitric oxide (NO) availability and actions (Patel et al., 2025; Reis et al., 2016; R. Roy et al., 2023). Still, whether NVC responses in the hypothalamus are altered in HF, thereby contributing to the exaggerated neurohormonal outflow in this condition, remains unexplored. Finally, dietary salt is a key factor contributing to volume overload and congestion in HF (Gupta et al., 2012; Mullens et al., 2024), and excessive dietary salt intake has been implicated in additional neurohumoral activation in both HF and hypertension (Appel, 2017; Gupta et al., 2012; Strazzullo et al., 2009). Thus, evaluating whether NVC responses to salt are altered in HF could shed light on alternative mechanisms by which salt could further contribute to complications in this prevalent cardiovascular disease. In this study, we used a combination of in vivo and ex vivo approaches to test the hypothesis that an imbalance in vasoconstrictive/vasodilatory signals in the hypothalamic SON leads to exacerbated salt‐evoked iNVC responses and consequently hypoxia in rats with HF. Materials and methods Ethical approval All performed experiments were approved by the Georgia State University (IACUC A25009) and the Augusta University Institutional Animal Care and Use Committee (IACUC 2011‐0319) and carried out in agreement with the IACUC guidelines. This study complied with the ethical principles and policies of The Journal of Physiology. Animals Male and female heterozygous transgenic eGFP‐AVP Wistar rats (250–400 g), originally kindly donated by Dr Ueta (Department of Physiology, School of Medicine, University of Occupational and Environmental Health, Japan) (Ueta et al., 2005). Subsequently, an in‐house breading colony of transgenic eGFP‐AVP Wistar rats were established and animals from this colony were used in all experiments. Rats were housed in cages (two per cage) under constant temperature (22 ± 2°C) and humidity (55 ± 5%) on a 12 h light cycle (lights on: 08.00–20.00 h). At all times, animals had ad libitum access to food and water, and every effort was made to minimize suffering and the number of animals used in this study. Heart failure surgery and echocardiography To induce HF in rats, a coronary artery ligation surgery was performed as previously described (Althammer et al., 2024a). Anaesthesia was induced using 5% isoflurane mixed with O2 (100% O2, 1 l/min). Optimum anaesthesia was assessed by the absence of limb withdrawal reflex to a painful stimulus (hind paw pinch). Rats were then intubated for mechanical ventilation. During the surgery, anaesthesia was adequately maintained using 2–3% isoflurane delivered by a vaporizer machine mixed with O2 (100% O2, 1 l/min). With aseptic precaution, a small incision was made in the third intercostal space, and a retractor was applied to retract the third and fourth ribs. The pericardium was ruptured, and the heart was exteriorized. A ‘Prolene 6‐0’ suture (Ethicone, USA) was inserted just beneath the left atrial base along the interventricular septum to make a loop around the main diagonal branch of the left anterior descending (LAD) coronary artery. Finally, myocardial infarction was induced by ligating the LAD coronary artery. Occlusion of the LAD coronary artery results in a pale discoloration of the left ventricle, which guides as a visual confirmation of the ischaemic left ventricle following LAD coronary artery ligation. Sham animals underwent a similar aseptic surgical procedure except for left coronary artery ligation. Four to five weeks after the surgery, we performed transthoracic echocardiography (Vevo 3100 systems; Visual Sonics, Toronto, ON, Canada) under light isoflurane (2–3%) anaesthesia (carried through 100% O2 at a flow rate of 1.5 ml/min) to assess the EF and confirm the development of HF. We obtained the left ventricle internal diameter and the left diameter of the ventricle posterior and anterior walls in the short‐axis motion imaging mode to calculate the EF. The myocardial infarct surgery typically results in a wide range of functional HF, as determined by the EF measurements. Rats with EF < 40% were considered as HF (sham 81.68 ± 2.75%, HF 28.54 ± 4.58; n = 20 and 25 in sham and HF rats, respectively, see Table 1 for additional echocardiographic parameters). Rats that underwent HF surgery but did not develop HF or displayed an EF > 40% were not included in the study. Table 1Summary of assessed cardiac parameters in sham and heart failure (HF) rats via echocardiography. Sham (n = 20) HF (n = 25) P value (unpaired t test) Heart rate (BPM)296.94 ± 27.22286.85 ± 19.82 P = 0.195Stroke volume (µl)231.66 ± 25.79137.28 ± 8.59 P < 0.001Ejection fraction (%)81.68 ± 2.7528.54 ± 4.58 P < 0.001Fractional shortening (%)48.25 ± 2.3115.48 ± 1.26 P < 0.001Cardiac output (ml/min)69.13 ± 6.5939.02 ± 3.61 P < 0.001LV mass (mg)1832.49 ± 71.69995 ± 51.29 P < 0.001LVAW d (mm)3.16 ± 0.241.38 ± 0.15 P < 0.001LVAW s (mm)3.82 ± 0.161.45 ± 0.16 P < 0.001LVID d (mm)6.87 ± 0.2412.38 ± 0.69 P < 0.001LVID s (mm)4.81 ± 0.388.14 ± 0.36 P < 0.001LVPW d (mm)4.10 ± 0.411.45 ± 0.25 P < 0.001LVPW s (mm)3.97 ± 0.161.96 ± 0.16 P < 0.001Abbreviations: LV, left ventricle; LVAW d, left ventricular end‐diastolic anterior wall thickness; LVAW s, left ventricular end‐systolic anterior wall thickness; LVPW d, left ventricular end‐diastolic posterior wall thickness; LVPW s, left ventricular end‐systolic posterior wall thickness; LVID d, left ventricular end‐diastolic diameter; LVID s, left ventricular end‐systolic diameter.John Wiley & Sons, Ltd. Summary of assessed cardiac parameters in sham and heart failure (HF) rats via echocardiography. Sham (n = 20) HF (n = 25) P value (unpaired t test) Abbreviations: LV, left ventricle; LVAW d, left ventricular end‐diastolic anterior wall thickness; LVAW s, left ventricular end‐systolic anterior wall thickness; LVPW d, left ventricular end‐diastolic posterior wall thickness; LVPW s, left ventricular end‐systolic posterior wall thickness; LVID d, left ventricular end‐diastolic diameter; LVID s, left ventricular end‐systolic diameter. Surgery for in vivo two‐photon imaging of the rat supraoptic nucleus A modified transpharyngeal surgical approach was used to expose the ventral surface of the brain containing the hypothalamic SON (Ludwig & Leng, 1997; R. K. Roy et al., 2021). Briefly, rats were anaesthetized by i.p. injection of urethane (1.5 g/kg) (U2500, Sigma Aldrich, MO, USA) or pentobarbitone (40 mg/kg, i.p.) when indicated (see Table 2). Upon cessation of the hindlimb withdrawal reflex, the trachea and left femoral vein were catheterized, and the rats were placed supine on a heating pad in a stereotaxic frame (Stoelting 51600 Lab Standard Stereotaxic Instrument). After the oral cavity was opened by splitting the mandibles, cautery (Thermal Cautery Unit, Geiger Medical Technologies, IA, USA) and drilling were used to remove the molars, ventromedial aspect of the right temporal bone, the basisphenoid and presphenoid bones to expose the ventral surface of the hypothalamus and ventral aspect of the right temporal lobe. The meninges overlying the right SON were removed, and a stainless‐steel ring (i.d. 3.6 mm) was placed on the surface of the brain with the junction of the internal carotid artery (ICA) and middle cerebral artery (MCA) at the centre. All the drugs used were topically delivered through an microdialysis (MD) probe (infusion rate 0.6 ml/h), which only penetrates a short distance inside the brain surface (Leng et al., 2001). To this end, a U‐shaped MD probe (permeable to 10 kDa; in‐house design) (Horn & Engelmann, 2001; Perkinson et al., 2022) was bent horizontally at an angle of 100–150° and placed on the exposed surface of the SON, where the PA, ICA and MCA were clearly visible inside the loop of the MD probe. In this process, drug concentrations achieved 0.5–1 mm below the brain surface are approximately four orders of magnitude lower than the concentration in the MD probe (Ludwig & Leng, 1997). Thus, to achieve a nanomolar range of drug concentration within the SON, drug concentration in the MD probe needs to be in the 1–2 mm range (Ludwig & Leng, 1997). The MD probe drug concentrations are noted in the corresponding results section. At the end of the experiments, rats were killed by decapitation using a guillotine. Table 2Summary of baseline plasma protein and osmolality obtained from urethane, pentobarbital and non‐anaesthetized control rats.UrethanePentobarbitalNon‐anaesthetized P value (one‐way ANOVA) Plasma osmolality (mOsmol/l) 318.91 ± 3.40 (n = 4) 317.66 ± 0.52 (n = 4) 319.58 ± 0.54 (n = 4) F (2, 10) = 0.34; P = 0.79 Bonferroni post hoc test Plasma protein (mg/ml) 38.86 ± 1.26 (n = 4) 38.99 ± 1.79 (n = 4) 38.25 ± 0.54 (n = 4) F (2, 10) = 0.73; P = 0.50 Bonferroni post hoc test John Wiley & Sons, Ltd. Summary of baseline plasma protein and osmolality obtained from urethane, pentobarbital and non‐anaesthetized control rats. P value (one‐way ANOVA) 318.91 ± 3.40 (n = 4) 317.66 ± 0.52 (n = 4) 319.58 ± 0.54 (n = 4) F (2, 10) = 0.34; P = 0.79 Bonferroni post hoc test Plasma protein (mg/ml) 38.86 ± 1.26 (n = 4) 38.99 ± 1.79 (n = 4) 38.25 ± 0.54 (n = 4) F (2, 10) = 0.73; P = 0.50 Bonferroni post hoc test In vivo two‐photon imaging of the SON and surrounding vasculature The microvasculature of the SON and the surrounding area was labelled by i.v. injection of rhodamine 70 kDa (20 mg/ml, 200 nl/rat) (R9379, Sigma Aldrich, MO, USA). The exposed ventral surface of the brain was covered with optical gel, and the SON and its surrounding vasculature were imaged under a two‐photon microscope (Bruker, Billerica, MA) excited with a Ti: Sapphire laser (Chameleon Ultra II (Coherent, Santa Clara, CA)) tuned at 840 or 940 nm and scanned with resonant galvanometers. As previously characterized (R. K. Roy et al., 2021), the visually accessible area of the SON is typically located medially to the bifurcation of the ICA and MCA. The SON was readily identified by the presence of eGFP‐AVP‐expressing neurons located within the dense capillary network. A schematic representation of the visually accessible area of the SON containing eGFP‐AVP neurons and surrounding vascular cytoarchitecture is shown in Fig. 1. For measurements of vessel diameter and blood flow, we used either a 4× (numerical aperture 0.13) or a 16× (numerical aperture 0.8) water immersion objective (Olympus, Centre Valley, PA, USA), respectively. Quantification of vessel diameter and blood flow Salt loading was performed via the femoral vein catheter by graded infusion of hypertonic saline (2 m NaCl, 1 ml over 30 min) in HF and control (sham) rats, which was not adjusted by body weight and was selected to match our previously established salt‐challenge paradigm for evoking SON neurovascular responses. Vessel diameter To quantify changes in vessel diameter during the salt loading challenge, a time‐series, Z‐stack acquisition protocol was used (one iteration: 20 images at 0.33 Hz, 30 µm interval up/down the Z‐axis in the SON; a total of 20 iterations with a 120 s interval; 512 × 512 pixels). Analyses were performed with ImageJ, as previously described (R. K. Roy et al., 2021). Briefly, a Z‐projected image was obtained for each of the 20 iterations, and vessel cross‐section lines were manually drawn perpendicular to the two sides of the vessel wall. A profile of image brightness along the cross‐section line of the vessel was then obtained. The high contrast between the vessel wall and the brain parenchyma allows for the ready detection of the vessel wall edges, which appear as a sudden rise in basal fluorescence. The distance between the two fast‐rising phases, measured just above the base of the fluorescence profile, was used to calculate the diameter in each frame of the time series. Plots of these measurements over the time series were obtained to study the time course of diameter changes during the salt challenge. Blood flow Blood flow was calculated by monitoring the movement of red blood cells (RBC) within the vessels, which appeared as moving dark spots in fluorescently labelled vessels through a segment of an imaged vessel. To this end, a separate set of experiments were performed, in which a line scan was performed at high speed (500–1000 Hz, 5–10 s) in a desired section of the targeted vessel. The line scans were readily converted into a kymographs by Prairie View software. In these graphs, the movement of RBCs appears as streaks in space‐time images, in which the streak angle is a function of the RBC velocity (Chhatbar & Kara, 2013; R. K. Roy et al., 2021). The kymograph generated from the line scan was then uploaded into Fiji software for the analysis of blood velocity (Schindelin et al., 2012). A line was manually drawn using the line tool along the streak angle to generate the angle of the line (the angle between the drawn line and the horizontal axis), taking care to consistently draw the line along the streak angle. This angle was then used to calculate the angle directly opposite the x‐axis (i.e. distance). After converting this angle to radians, now represented as ⍺, the following equation was used to calculate velocity: (1) Velocity=tanα Once the velocity ratio was calculated, it was converted into µm/ms using the embedded distance (µm/pixel) and line period (ms/pixel) values from the line scan. This same procedure was repeated 20 times for each kymograph, sampling a different band each time to account for RBC velocity variability within the sampled time. The velocity values were then averaged for every line scan at each time point. Measurement of tissue partial pressure of oxygen (PO2) The partial pressure of tissue O2 (PO2) in the SON was measured using an optical fluorescence probe (250 µm tip diameter; OxyLite system; Oxford Optronix), which allows real‐time recording of absolute changes in tissue PO2 as previously described. Rats were anaesthetized with urethane (1.5 g/kg, i.p.) and positioned in a stereotaxic apparatus. Adequate depth of anaesthesia was confirmed by the absence of a withdrawal reflex in response to a firm toe pinch. The SON was exposed using the transpharyngeal preparation described above. For these experiments, rats were mechanically ventilated to obtain a stable baseline PO2. The sensor was advanced under visual guidance using the bifurcation of the ICA and MCA, together with the previously characterized vascular landmarks identifying the visually accessible SON region. The probe tip was positioned approximately 1–2 mm below the ventral surface of the brain, within the SON parenchyma, then slightly retracted (∼0.1 mm) to minimize tissue compression and allowed to equilibrate for 15–20 min before recording. Because of the size of the optical probe, these measurements were intended to estimate local PO2 within the SON tissue and not the precise oxygen microenvironment surrounding an individual imaged arteriole. At the end of the experiments, rats were killed by decapitation using a guillotine. Measurement of plasma osmolality and plasma protein levels Blood was collected from the femoral vein after the salt challenge (2 m NaCl, 1 ml i.v. over 30 min) or an isotonic saline infusion in sham and HF rats. Plasma was separated via centrifugation (1000‐2000 g). The osmolality of the freshly obtained plasma was measured using a micro‐osmometer (Advanced Instruments, Model 3300, Version 4.3, Massachusetts, USA). Protein concentration was measured from freshly obtained plasma samples using a Nanodrop Ultra spectrophotometer (Thermo Fisher Scientific, USA). Quantitative measurement of local AVP release within the SON using biosensor SnifferAVP cells The full description of this methodological approach has been previously described (Pitra et al., 2019). Briefly, SnifferAVP biosensors were generated by culturing Chinese hamster ovary cells in Dulbecco's Modified Eagle Medium containing 10% w/v fetal bovine serum, 1% w/v penicillin–streptomycin, 1% w/v Na‐pyruvate and 1% w/v NaHCO3 filtered once through a Nalgene filtration system, and transfected with pcDNA3.1+ containing human V1A vasopressin receptors cloned in EcoRI (5’) and XhoI (3’) (plasmid obtained from Missouri S&T cDNA Resource Centre, Rolla, MO, USA) using lipofectamine, and stable over expression was achieved by geneticin (500 mg/ml) selection. About 16−20 h before the experiment, SnifferAVP cells were transiently transfected to express the red fluorescent genetically encoded calcium indicator R‐GECO (GenScript, Piscataway, NJ, USA) with Fugene HD reagent (Promega, Madison, WI, USA). SnifferAVP cells were resuspended in standard aCSF (in mm): 119 NaCl, 2.5 KCl, 1 MgSO4, 26 NaHCO3, 1.25 NaH2PO4, 20 d‐glucose, 0.4 ascorbic acid, 2 CaCl2 and 2 pyruvic acid; pH 7.3; 300−305 mOsm) with trypsin (0.05%). Coronal brain slices (300 µm thick) were obtained from rat brains following administration of a lethal dose of pentobarbital sodium (50 mg/kg, i.p.) using a vibratome (VT1200S, Leica Microsystems, Germany). SnifferAVP biosensor cells were transferred via pipette directly onto the SON of coronal brain slices submerged in aCSF in a perfusion‐capable chamber of an upright microscope. For anatomical precision, the optic tract was used as a landmark. Before all recordings, eGFP‐AVP neurons within the SON were observed in the field of view. An example of a SON slice containing both eGFP‐AVP neurons and the SnifferAVP biosensors is shown in Fig. 6. After allowing the SnifferAVP biosensors to adhere to the slice for 5 min, aCSF superfusion was resumed to wash off any unattached SnifferAVP from the slice before proceeding with imaging. Experiments were restricted to preparations that had at least five fluorescently visible sniffer cells in the field (∼10 on average). To record the calcium‐induced fluorescence changes of the SnifferAVP biosensors, images were captured using a Dragonfly 200 confocal scanning system and an iXON EMCCD camera (Andor Systems, Belfast, UK), at a rate of 1 Hz, with piezo‐driven z‐series to maximize biosensor counts per trial. Slices were constantly superfused with aCSF at 32°C at ∼2 ml/min. Imaging data were analysed using ImageJ software (NIH). All data were background‐subtracted. For quantitative measurements, fractional fluorescence (F/F0) was determined by dividing the fluorescence intensity (F) within a region of interest by the baseline fluorescence value (F0), which was determined from 30 frames preceding stimulation. Peak sniffer Ca2+ amplitude was the maximum F/F0 achieved following UV laser uncaging of MNI‐caged‐l‐glutamate (25 µm) in the presence of DNQX (10 µm) to restrict activation of AMPA receptors. As shown in Fig. 6, glutamate uncaging resulted in a robust burst firing in AVP neurons, which we have previously shown to efficiently evoke local release of AVP within this SON (Pitra et al., 2019; Son et al., 2013). The area under the curve (F/F0∗s) was calculated by integrating the response over its duration. Response rates represent the number of biosensors that responded to the stimulation, divided by the total number of biosensors present in the field of view. SnifferAVP biosensors that showed intrinsic oscillatory calcium activity were excluded from the analysis. To better display changes in fluorescence levels, images were pseudo‐coloured using ImageJ. Statistical analysis All statistical analyses were performed using GraphPad Prism 10 (GraphPad Software, CA, USA). Both male and female rats were included in the study. However, the experiments were not designed or statistically powered to assess sex as an independent biological variable; therefore, formal sex‐based comparisons were not performed, and no conclusions regarding sex differences are drawn from these data. The significance of differences was determined using paired or unpaired t tests (two‐sided in all cases), one‐way or two‐way repeated‐measures (RM) or normal analysis of variance (ANOVA) with the post hoc Bonferroni multiple comparisons test. Results are expressed as means ± standard deviation (SD). Results were considered to be statistically significant if P < 0.05 and are presented as ∗ for P < 0.05, ∗∗ for P < 0.01 and ∗∗∗ for P < 0.0001 in the respective figures. Results Sustained hypoxic milieu in the SON of HF rats A schematic of the two‐photon in vivo hypothalamic preparation is shown in Fig. 1A . Similar to our previous angioarchitecture characterization (R. K. Roy et al., 2021), the SON in HF rats was readily identified by the presence of eGFP‐AVP‐expressing neurons which reside within the dense capillary network (labelled with i.v. injection of rhodamine 70 kDa dextran dye, Fig. 1B ). This dense capillary network is located medially to the bifurcation of the ICA, one of the main pial vessels supplying the SON (Ambach & Palkovits, 1979) and the branching of the MCA from the ICA (Fig. 1B ). As previously described, parenchymal arterioles (PAs) originated near the most caudal aspect of the SON, which subsequently branched into smaller arterioles, forming this dense capillary network. We first determined whether, in sham versus HF, baseline vessel diameter differed across main SON arterioles (i.e. PA, ICA, MCA). In vivo two‐photon imaging of the SON vasculature revealed no difference in the different vessel types (i.e. PA, ICA and MCA) between sham and HF rats (F = 3.2, P = 0.08, two‐way ANOVA, n = 6 of each vessel/6 rats in each group) (Fig. 1C ). However, a significant decrease in basal blood flow was observed in PAs of HF compared with sham rats (sham: 30.01 ± 1.78 µm/ms; HF: 20.27 ± 3.18 µm/ms; n = 3 in sham and 4 in HF; P = 0.007; unpaired t test) (Fig. 1D ). We next assessed whether SON parenchymal PO2 saturation level was altered during HF. To this end, we placed an optical fluorescence probe (250 mm tip diameter; OxyLite system; Oxford Optronix) in the SON parenchyma, which allows real‐time recording of parenchymal PO2 level (R. K. Roy et al., 2021). We found that SON PO2 saturation was lowered in HF compared with sham rats (sham: 22.87 ± 4.16 mmHg; HF: 16.66 ± 2.73 mmHg; n = 6; P = 0.019; unpaired t test) (Fig. 1E ). Taken together, these results support sustained tissue hypoxia within the SON of HF rats at baseline conditions. A persistent hypoxic milieu in the supraoptic nucleus (SON) of heart failure (HF) rats A, schematic representation of the preparation used for in vivo two‐photon imaging of the SON. Note the presence of the microdialysis probe positioned right on top of the SON, used to locally deliver drugs into the region. B, low‐magnification (16×) two‐photon image of the visually accessible area of the SON depicting main vessels in red following rhodamine 70 kDa administration (i.v.). The SON area is identified by the presence of a dense capillary network (red) along with eGFP‐AVP‐positive neurons (green), medially to the bifurcation of the internal carotid artery (ICA) and the middle cerebral artery (MCA). The magnified image to the right shows the capillary network and vasopressin (AVP) cells from the area bounded by the white rectangle in the low‐magnification image to the left. C, bar graph summarizing the mean basal diameter of the parenchymal arteriole (PA), ICA, and the MCA in sham and HF rats (n = 6 of each vessel/6 rats in each group). D, bar graph summarizing the mean blood velocity in SON PA of sham and HF rats (n = 3 in sham and n = 4 in HF group, **P = 0.007; unpaired t test). E, bar graph representing the mean oxygen saturation level (PO2) in the SON of sham and HF rats (n = 6 in each group, *P = 0.019; unpaired t test). Scale bar = 100 µm for low magnification and 25 µm for high magnification. Error bars represent SDs. Reversed polarity of the salt‐evoked NVC in HF rats Our previous work demonstrated that a systemic salt challenge, a potent stimulation of SON AVP neuronal activity, evoked a vasoconstriction of SON PAs (i.e. iNVC) that was mediated by somatodendritic release of AVP (R. K. Roy et al., 2021). Given that AVP expression and release are both enhanced in HF rats (Chen et al., 2015; Kim et al., 1990; Reis et al., 2016), we hypothesized an amplified iNVC response in this experimental group. To test this hypothesis, sham and HF rats were subjected to a standardized hypertonic saline challenge (2 m NaCl, 2 ml/h, intravenous infusion for 30 min). At the same time, continuous measurements of PA diameter were obtained using two‐photon imaging. In prior work using this same protocol, hypertonic saline infusion did not alter any major systemic cardiovascular or respiratory parameters, including mean arterial pressure, heart rate, blood gases (PO2, PCO2, HCO3 −), or arterial pH (R. K. Roy et al., 2021). To rule out any potential effects of urethane anaesthesia on plasma osmolality and volaemia, baseline measurements of these parameters were obtained in rats anaesthetized with either urethane or pentobarbital and compared with values from non‐anaesthetized rats. As summarized in Table 2, plasma osmolality and plasma protein concentration did not differ among groups. As previously reported, we found that this salt challenge evoked a change in plasma osmolality of ∼40 mOsm/l, with no differences observed between sham and HF (sham: 361 ± 0.81 mOsm/l; HF: 359 ± 1.41 mOsm/l, n = 4 in sham and n = 5 in HF: P > 0.05; unpaired t test) (Table 3). Representative two‐photon images of SON PAs before and during the salt challenge in sham and HF rats are shown in Fig. 2Aa–Ae and Fig. 2Ba–Be , respectively. Consistent with our previous findings obtained in wild‐type rats, sham rats in this study showed vasoconstriction of SON PAs in response to acute salt loading (one‐way RM ANOVA; F = 37.5, P < 0.001; n = 6 vessels/6 rats) (Fig. 2C ). In contrast, the same procedure in HF rats evoked a pronounced vasodilatation of SON PAs (one‐way RM ANOVA; F = 102.8 P < 0.001; n = 6 vessels/6 rats) (Fig. 2C ). No noticeable differences in the onset of the vasoconstriction or vasodilatation were observed, in both cases starting at ∼10 min after the salt infusion was initiated. Furthermore, vascular responses persisted in both groups throughout the salt challenge. Table 3Summary of plasma osmolality in control (no salt loading), sham and heart failure (HF) rats after salt loading.ControlShamHF P value (unpaired t test) Plasma osmolality (mOsmol/l) 316.95 ± 4.05 (n = 7) 361 ± 0.81 (n = 4) 359 ±1.41 (n = 5) P = 0.24 (sham vs. HF) P < 0.001 (cont vs. sham) P < 0.001 (cont vs. HF) John Wiley & Sons, Ltd. Summary of plasma osmolality in control (no salt loading), sham and heart failure (HF) rats after salt loading. P value (unpaired t test) 316.95 ± 4.05 (n = 7) 361 ± 0.81 (n = 4) 359 ±1.41 (n = 5) P = 0.24 (sham vs. HF) P < 0.001 (cont vs. sham) P < 0.001 (cont vs. HF) The polarity of the salt‐evoked neurovascular coupling response is reversed in heart failure (HF) rats A, low‐magnification in vivo two‐photon images of the supraoptic nucleus (SON) following rhodamine 70 kDa administration (i.v.) before (Aa), and during (Ab), the salt challenge (2 m NaCl, 1 ml, 2 ml/h) in a representative sham rat. The blue asterisk indicates the shadow cast by the microdialysis probe. Ac and Ad, the magnified image of the parenchymal arteriole (PA) (demarcated as a red asterisk) in Aa and Ab, respectively. In Ae, the profiles of image brightness along the cross‐section line of the PA vessels shown in Ac (before, black) and during (Ad, red) the salt challenge are shown. Note the reduced PA diameter (vasoconstriction) during the salt challenge. B, low‐magnification in vivo two‐photon images of the SON following rhodamine 70 kDa administration (i.v.) before (Ba), and during (Bb), the salt challenge (2 m NaCl, 1 ml, 2 ml/h) in a representative HF rat. Bc and Bd, the magnified image of the PA (demarcated as a red asterisk) in Ba and Bb, respectively. In Be, the profiles of image brightness along the cross‐section line of the PA vessels, as shown in Bc (before, black) and during (Bd, red) the salt challenge, are displayed. Note the increased PA diameter (vasodilatation) during the salt challenge. C, summary plot of the mean percentage change in PA diameter as a function of time during the salt challenge in sham (blue) and HF (red) rats. Note the significant vasoconstriction evoked in sham rats (blue, F = 37.5, P < 0.001, one‐way RM ANOVA) and the vasodilatation evoked in HF rats (red, F = 102.8, P < 0.001, one‐way RM ANOVA, n = 6 vessels/6 rats in each experimental group). D, representative raw traces showing changes in SON tissue partial pressure of oxygen (PO2) before and during the salt infusion in a representative sham (upper) and HF (lower) rat. E, bar graph summarizing the mean SON PO2 before and 30 min after the salt challenge in sham (blue) and HF (red) rats. n = 6 rats in each group. Scale bar = 200 µm. Error bars represent SDs. *P = 0.002 for sham basline vs. sham 30 min post‐salt, *P = 0.003 for HF baseline vs. HF 30 min post‐salt, one‐way ANOVA, Bonferroni post hoc test. To determine whether the salt‐evoked NVC responses impacted the PO2 within the SON, we measured PO2 saturation levels during the salt challenge in both experimental groups. Representative traces showing the change in PO2 in the SON of a sham and HF rats are shown in Fig. 2D . Consistent with the vasoconstriction response evoked in sham rats, the acute salt challenge significantly reduced SON PO2 saturation SON in sham rats (baseline: 22.99 ± 4.35 mmHg; after salt loading: 15.47± 3.75 mmHg; n = 6; P = 0.022; unpaired t test) (Fig. 2E ). Conversely, but consistent with the vasodilatory response in HF rats, the acute salt challenge significantly increased the PO2 saturation in the SON of HF rats (baseline: 18.84 ± 5.19 mmHg; after salt loading: 25.93 ± 5.01 mmHg; n = 7; P = 0.033; Bonferroni post hoc test), normalizing the baseline hypoxic condition, to similar baseline levels observed in sham rats (Fig. 2E ). Taken together, these results support a change in the polarity of the vascular response (from vasoconstriction to vasodilatation) in the salt‐evoked NVC response in HF rats, which effectively resulted in an increase and normalization of the PO2 saturation level within the SON. The salt‐evoked vasodilatation in HF rats is mediated by adenosine acting on A2AR receptors Previously, we demonstrated that the salt‐evoked vasoconstriction in the SON was mediated via activity‐dependent, local dendritic release of AVP (R. K. Roy et al., 2021). In HF rats, microdialysis of AVP directly onto the SON (1 mm, 2.2 ml/h for 5 min) evoked a progressive vasoconstriction of SON PAs (one‐way RM ANOVA, F = 20.8, P = 0.005, n = 5 vessels/5 rats) (Fig. 3A ). Thus, given that salt loading elicited vasodilatation rather than vasoconstriction in HF rats, these results indicate that AVP retains its vasoconstrictive action in HF and is therefore unlikely to directly mediate the net salt‐evoked vasodilatation observed in this group. We therefore considered the possibility that salt loading in HF recruits an additional vasodilatory signal that predominates over, or masks, the AVP‐dependent vasoconstrictive influence. A potential candidate is adenosine. Adenosine, acting on adenosine A2A receptors (A2AR), causes potent vasodilatation and can contribute to NVC responses in the somatosensory cortex and retina. To investigate whether a similar signalling cascade could contribute to the salt‐evoked vasodilatation in HF rats, we first tested whether microdialysis of CGS21680, a selective A2AR agonist, directly onto the SON of HF rats (1 mm, 2.2 ml/h for 5 min) induced a vasodilatory response. As shown in Fig. 3B , CGS21680 caused a progressive vasodilatation of SON PAs in HF rats (one‐way RM ANOVA, F = 21.0, n = 4 vessels/4 rats; P = 0.011). Notably, infusion of the A2AR antagonist, the vasopressin V1A receptor (V1AR) antagonist, or aCSF (vehicle) alone did not alter baseline vascular diameter (one‐way RM ANOVA: F = 5.52, P = 0.081; n = 5 vessels from five rats for V1AR antagonist and n = 4 vessels from four rats for A2AR antagonist and aCSF) (Fig. 3C ). Vasopressin (AVP) and adenosine evoke vasoconstriction and vasodilatation, respectively, in supraoptic nucleus (SON) parenchymal arterioles (PAs) in heart failure (HF) rats A, low‐magnification in vivo two‐photon images of the SON following rhodamine 70 kDa administration (i.v.) before (Aa), and during (Ab), topical infusion of AVP (1 mm, 2.2 ml/h) through a microdialysis probe positioned on top of the SON in a representative HF rat (the blue asterisk indicates the shadow cast by the microdialysis probe). Ac and Ad, the magnified image of the PA (demarcated as a red asterisk) in Aa and Ab, respectively. In Ae, the profiles of image brightness along the cross‐section line of the PA vessels shown in Ac (before, black) and in Ad (during, red), the AVP infusion, are displayed. Note the reduced PA diameter (vasoconstriction) evoked by AVP. Af, bar graph showing the mean changes in PA diameter after the local infusion of AVP. n = 5 vessels/5 rats; **P = 0.005, paired t test. B, low‐magnification in vivo two‐photon images of the SON following rhodamine 70 kDa administration (i.v.) before (Ba), and during (Bb), topical infusion of the adenosine A2A receptor agonist CGS 21680 (1 mm, 2.2 ml/h) through a microdialysis probe positioned on top of the SON in a representative HF rat (the blue asterisk indicates the shadow cast by the microdialysis probe). Bc and Bd, the magnified image of the PA (demarcated as a red asterisk) in Ba and Bb, respectively. In Be, the profiles of image brightness along the cross‐section line of the PA vessels shown in Bc (before, black) and in Bd (during, red), the CGS 21680 infusion, are displayed. Note the increased PA diameter (vasodilatation) evoked by CGS 21680. Bf, bar graph showing the mean changes in PA diameter after the local infusion of CGS 21680. n = 3 vessels/3 rats; *P = 0.011, paired t test. C, bar graph representing the PA vessel diameter before and after microdialysis probe infusion of an A2AR antagonist (blue, ZM 241385), a V1AR antagonist (red, β‐mercapto‐β,‐cyclopentamethylenepropionyl1), or aCSF (vehicle) (cyan). A2AR antagonist: P = 0.83 (n = 4); V1AR antagonist: P = 0.85 (n = 5); and aCSF: P = 0.99 (n = 4), paired t test in all cases. Scale bar = 200 µm. Error bars represent SDs. These results indicate that A2AR activation is sufficient to dilate SON PAs in HF rats, supporting adenosine/A2AR signalling as a candidate mechanism contributing to the salt‐evoked vasodilatory response in this condition. To test this, we performed a separate set of experiments in which HF rats were subjected to a systemic salt challenge (2 m NaCl, IV, 1 ml/h) while ZM 241385, a selective A2AR antagonist, was microdialysed directly into the SON (1 mm, 0.6 ml/h). As shown in Fig. 4Aa–Ae , local delivery of an A2AR antagonist into the SON not only prevented salt‐induced vasodilatation but unmasked an underlying vasoconstriction (one‐way RM ANOVA: F = 35.9, P < 0.001; n = 6 vessels from six rats; Fig. 4D ). This unmasked vasoconstriction was abolished when the A2AR antagonist was co‐infused with a V1AR antagonist (β‐mercapto‐β,‐cyclopentamethylenepropionyl1, [O‐me‐Tyr2, Arg8]‐AVP; 200 µm) (Fig. 4Ba–Be,D ; one‐way RM ANOVA: F = 1.4, P = 0.6; n = 4 vessels from four rats). The salt‐evoked vasodilatory response in heart failure (HF) rats is mediated by endogenously released adenosine acting on A2A receptors A, low‐magnification in vivo two‐photon images of the supraoptic nucleus (SON) following rhodamine 70 kDa administration (i.v.) before (Aa), and during (Ab), the salt challenge (2 m NaCl, 1 ml, 2 ml/h) in the presence of microdialysed ZM 241385 (a selective A2AR antagonist,1 mm, 0.6 ml/h) in a representative HF rat. The blue asterisk indicates the shadow cast by the microdialysis probe. Ac and Ad, the magnified image of the parenchymal arteriole (PA) (demarcated as a red asterisk) in Aa and Ab, respectively. In Ae, the profiles of image brightness along the cross‐section line of the PA vessels shown in Ac (before, black) and during (Ad, red) the salt challenge, are displayed. Note the reduced PA diameter (vasoconstriction) during the salt challenge in the presence of the A2AR blocker. B, low‐magnification in vivo two‐photon images of the SON following rhodamine 70 kDa administration (i.v.) before (Ba), and during (Bb), the salt challenge in the presence of co‐microdialysed (0.6 ml/h) ZM 241385 (1 µm) plus β‐mercapto‐β,‐cyclopentamethylenepropionyl (a V1A receptor antagonist, 200 µm) in another representative HF rat. Bc and Bd, the magnified image of the PA (demarcated as a red asterisk) in Ba and Bb, respectively. In Be, the profiles of image brightness along the cross‐section line of the PA vessels shown in Bc (before, black) and during (Bd, red) the salt challenge, are displayed. Note lack of changes PA diameter during the salt challenge in the presence of both A2AR and V1A receptor blockers. C, low‐magnification in vivo two‐photon images of the SON following rhodamine 70 kDa administration (i.v.) before (Ca), and during (Cb), the salt challenge in the presence of microdialysed PSB 0739 (a selective purinergic receptor P2Y12 antagonist, 1 µm 0.6 ml/h) in another representative HF rat. Cc and Cd, the magnified image of the PA (demarcated as a red asterisk) in Ca and Cb, respectively. In Ce, the profiles of image brightness along the cross‐section line of the PA vessels shown in Cc (before, black) and during (Cd, red) the salt challenge, are displayed. Note the reduced PA diameter (vasoconstriction) during the salt challenge in the presence of the P2Y12 blocker. D, summary plot of the mean percentage change in PA diameter as a function of time during the salt challenge in HF rats in the control conditions (red), or in the presence of microdialysed A2AR antagonist (green), P2Y12R antagonist (black), or a combination of a V1AR antagonist with the A2AR antagonist (purple). n = 5 vessels/5 rats in each group. *P = 0.03, **P = 0.002, ***P < 0.001, control group. #P = 0.03, ##P = 0.002, ###P <0.001, for A2AR antagonist group, ^P = 0.011, ^^P = 0.004, ^^^P < 0.001, P2Y12R antagonist group, one‐way ANOVA followed by Bonferroni post hoc test, vs. baseline. E, bar graph showing the mean peak magnitude of the salt‐evoked vasoconstriction in sham rats (blue, n = 6 vessels/6 rats), and HF rats in the presence of the A2AR antagonist (pink, n = 6 vessels/6 rats) and the P2Y12R antagonist (black, n = 5 vessels/5 rats). One‐way ANOVA, F = 11.3, P < 0.001, Bonferroni post hoc test. Scale bar = 200 µm. Error bars represent SDs. To further determine whether the salt‐evoked A2AR‐mediated vasodilatation occurred independently of AVP signalling or required intact AVP/V1AR signalling, we repeated the systemic salt challenge during continuous local blockade of V1A receptors alone. Under these conditions, salt infusion failed to induce any detectable change in vascular diameter, with neither vasoconstriction nor vasodilatation observed (one‐way RM ANOVA: F = 1.68, P = 0.4; n = 5 vessels from five rats; Fig. 5). This result argues against a model in which salt independently activates a fully AVP‐independent adenosine‐mediated vasodilatory pathway. Instead, it supports the interpretation that intact AVP/V1AR signalling is required, either upstream of, or permissively, for the adenosine/A2AR‐mediated vasodilatory response to emerge during salt loading in HF rats. The salt‐evoked, adenosine‐mediated vasodilatation in heart failure (HF) rats is dependent on an intrinsic vasopressin (AVP) signalling mechanism A, low‐magnification in vivo two‐photon images of the supraoptic nucleus (SON) following rhodamine 70 kDa administration (i.v.) before (Aa), and during (Ab), the salt challenge (2 m NaCl, 1 ml, 2 ml/h) in the presence of microdialysed V1A receptor blocker β‐mercapto‐β,‐cyclopentamethylenepropionyl (200 µm, 0.6 ml/h) in a representative HF rat. The blue asterisk indicates the shadow cast by the microdialysis probe. Ac and Ad, the magnified image of the parenchymal arteriole (PA) (demarcated as a red asterisk) in A1 and A2, respectively. B, the profiles of image brightness along the cross‐section line of the PA vessels shown in Ac (before, black) and during (Ad, red) the salt challenge are shown. Note the lack of changes in PA diameter during the salt challenge in the presence of the V1AR blocker. C, summary plot of the percentage change in PA diameter as a function of time during the salt challenge in HF rats in the presence of microdialysed V1AR antagonist (green). n = 5 vessels/5 rats in each condition; F = 0.76, P = 0.69, one‐way ANOVA. Scale bar = 200 µm. Error bars represent SDs. Together, these results indicate that salt loading in HF rats recruits opposing vasoactive influences within the SON: an AVP/V1AR‐dependent vasoconstrictive component and an adenosine/A2AR‐mediated vasodilatory component. The absence of both vasoconstriction and vasodilatation during V1AR blockade alone further suggests that the net vasodilatory response is functionally dependent on intact AVP/V1AR signalling, rather than reflecting a fully independent adenosine pathway. Microglia contribute to adenosine‐mediated, salt‐evoked vasodilatory responses in HF rats In the central nervous system (CNS), microglia constitute an important source of adenosine (Jackson et al., 2017; Marciante et al., 2024), and recent studies indicate that microglia actively regulate CBF via an adenosine signalling mechanism (Császár et al., 2022b; Fu et al., 2025). Moreover, the purinergic receptor P2Y12 is exclusively expressed on microglia (Lin et al., 2021; Zhu et al., 2017), activation of which plays a pivotal role in microglial activation and the release of signals in response to various stimuli (Light et al., 2006; Swiatkowski et al., 2016; Yu et al., 2019). To investigate whether microglia could be a potential cellular source contributing to the salt‐evoked vasodilatation of SON PAs in HF rats, we repeated a systemic salt challenge in a separate set of experiments while PSB0739, a selective P2Y12 purinergic receptor antagonist, was microdialysed directly into the SON (1 µm, 0.6 ml/h). Similarly to what we observed with the A2AR antagonist, PSB0739 not only prevented the salt‐evoked vasodilatation but also unmasked an underlying vasoconstriction in HF rats (one‐way RM ANOVA; F = 77.8, P < 0.001; n = 5 vessels/5 rats) (Fig. 4Ca‐Ce,D ). Although we did not directly test whether the vasoconstriction unmasked by P2Y12 receptor blockade was prevented by V1AR blockade, the similarity between this response and the vasoconstriction unmasked by A2AR blockade, together with the ability of V1AR blockade to abolish salt‐evoked vascular responses in HF rats, suggests that P2Y12 receptor blockade reveals the same underlying AVP/V1AR ‐dependent vasoconstrictive component. Taken together, these results suggest that microglia may present one potential source of adenosine contributing to salt‐evoked vasodilatation of SON PAs in HF rats. An increased activity‐dependent somatodendritic release of AVP likely contributes to an enhanced AVP‐mediated vasoconstriction (iNVC) in HF rats Notably, we found that the magnitude of the ‘masked’ salt‐evoked, AVP‐mediated vasoconstriction in HF rats (i.e. in the presence of the A2AR or the P2Y12R antagonists) was significantly larger than that evoked in sham rats (HF‐A2AR antagonist: ‐29.0% ± 5.37%, HF‐ P2Y12R antagonist: ‐28.12% ± 6.44%; sham: ‐11.62% ± 3.67%; n = 6 vessels/6 rats in each group; one‐way ANOVA, F = 11.3, P < 0.001; unpaired t test) (Fig. 4E ). These results indicate that even though the salt‐evoked adenosine‐mediated vasodilatation overrides the AVP‐mediated vasoconstriction in HF rats, somatodendritic release of AVP may still be enhanced in this condition. To test whether this was the case, we used SnifferAVP biosensor cells (see Methods), an approach we and others recently demonstrated can detect endogenous release of neuropeptides with extremely high sensitivity and spatiotemporal resolution (Perkinson et al., 2022; Pitra et al., 2019; Zaelzer et al., 2018). SnifferAVP biosensors were plated into ex vivo slices containing the SON, obtained from either sham or HF rats (n = 4 and 5 sham and HF rats, respectively). Photolysis of caged glutamate was used to stimulate NMDA receptor‐evoked firing and somatodendritic release of AVP within the SON (see Methods), an approach we previously demonstrated to be highly efficient in evoking this local release modality (Pitra et al., 2019). To validate the photolysis approach, we first obtained patch clamp recordings from identified AVP neurons under control conditions, demonstrating that our laser ‘uncaging’ protocol worked as intended. As shown in Fig. 6, laser photolysis of caged glutamate (UV flash, 2 s) induced a robust burst of action potential in patched AVP neurons (mean number of action potentials, 38.3 ± 6.4, mean interspike frequency: 22.8 ± 3.1 Hz (significantly higher than baseline frequency of 3.4 ± 0.9 Hz, P < 0.001, paired t test, n = 8). As previously reported, we found that NMDA‐evoked firing in AVP neurons resulted in local AVP release, as shown by the responsiveness of SnifferAVP biosensors (i.e. increased in Ca2+ fluorescence). Importantly, we found that the percentage of responsive SnifferAVP biosensors was significantly higher in slices obtained from HF (∼57%, 38/67) compared with sham rats (∼37%, 46/124, P < 0.05, unpaired t test, Fig. 6). Conversely, the peak and area of the Ca2+ response in the responsive SnifferAVP biosensors was similar between the two experimental groups (P = 0.7 and 0.1, for peak and area, respectively, unpaired t test, n = 46 and 38 SnifferAVP biosensors from sham and HF rats, respectively, Fig. 6). Together, these results support an increased activity‐dependent somatodendritic release of AVP in HF rats. Enhanced activity‐dependent local release of AVP within the supraoptic nucleus (SON) of heart failure (HF) rats A, representative image of the SON in a hypothalamic slice containing eGFP‐AVP neurons (green) and the sniffer‐AVP biosensor cells (yellow) plated on the SON. B, representative example of a patched AVP‐GFP neuron showing a robust burst of action potentials evoked by focal glutamate uncaging (UV light, 2 s, black bar). C, pseudocolour Ca2+ images showing responsive sniffer‐AVP cells (arrows) at baseline (Ca), immediately after stimulation (Cb) and the recovery after stimulation (Cc), following glutamate uncaging in a representative slice obtained from an HF rat. The inset in Cb shows corresponding Ca2+ traces of the sniffer‐AVP cells indicated with the corresponding asterisks. D, bar graphs showing mean data of the percentage of responsive sniffer‐AVP cells (Da), Ca2+ peak (Db), and Ca2+ area (Dc), in sham and HF rats. *P < 0.05 (unpaired t test). Error bars represent SDs. Discussion Neurovascular coupling (NVC) is a critical brain process that ensures increased CBF in response to neuronal activity (Iadecola, 2017; Urban et al., 2017). Recently, an impaired NVC response in the cortex has been reported both in animal models (Hoyer‐Kimura et al., 2024) and human patients (Aires et al., 2020) with HF, a pathophysiological mechanism proposed to contribute to cognitive impairment and dementia in this prevalent cardiovascular disease. Most existing studies on NVC, however, are limited mainly to superficial dorsal brain areas, particularly the cortex, and focus on responses to specialized sensory inputs. In these cases, neuronal activity‐evoked CBF changes occur rapidly (<2 s) and in a spatially restricted manner (Silva et al., 2000). Recently, we addressed this limitation by implementing a novel in vivo, real‐time two‐photon experimental approach to assess NVC for the first time in deep brain hypothalamic nuclei, including the SON (R. K. Roy et al., 2021; R. K. Roy et al., 2024). In this recent work, we demonstrated that as opposed to the conventional functional hyperaemic response previously described, increased activity of SON AVP neurons via a systemic salt challenge evoked PA vasoconstriction (hence termed inverse NVC (iNVC)) and a progressive decrease in SON PO2. Moreover, we demonstrated that iNVC was mediated by local somatodendritic release of AVP within the SON (R. K. Roy et al., 2021). The hypothalamus is a critical brain region for the generation of homeostatic neurohumoral responses (Saper & Lowell, 2014). Moreover, a large body of evidence supports a hypothalamic contribution to exacerbated neurohumoral responses in HF, resulting in higher morbidity and mortality in this disease (Cohn, 1996; Hartupee & Mann, 2017; Manolis et al., 2023). Given that excess dietary salt triggers neurohumoral responses involving sympathoexcitation and the release of AVP (Antunes‐Rodrigues et al., 2004; Bourque, 2008), and that it worsens fluid retention and congestion in HF (Gupta et al., 2012), we considered it imperative to determine whether salt‐evoked NVC responses were altered in HF, in a way that could contribute to the pathophysiology of this disease. Nitric oxide (NO) is a major systemic and central vasodilator (Ahmad et al., 2018), and a blunted hypothalamic NO expression and function has been reported in HF rats (Patel et al., 2025; Reis et al., 2016; R. Roy et al., 2023). Conversely, the expression and release of AVP, a potent vasoconstrictor, are elevated in HF rats (Patel et al., 2000; Price et al., 2004). Based on the imbalance of these two major vasoactive modulators favouring vasoconstriction in HF, and the fact that we showed salt‐evoked iNVC to be mediated by AVP (R. K. Roy et al., 2021), we hypothesized that exacerbated salt‐evoked vasoconstrictions would occur in the SON of HF rats, leading to a more pronounced hypoxic milieu in this condition. Surprisingly, we observed quite the opposite response. Our main results can be summarized as follows: (1) under baseline conditions, HF rats exhibit lower levels of PO2 in SON; (2) Despite this initial basal SON hypoxic condition, a salt challenge induced parenchymal arteriolar vasodilatation and a progressive increase in PO2, restoring it to levels comparable to those observed in sham rats; (3) The salt‐evoked vasodilatation in HF was mediated by adenosine, acting on adenosine A2A receptors, of a putative microglial source; (4) A masked AVP‐mediated vasoconstriction was still observed, which was of an even larger magnitude than that observed in sham rats, and was likely due to an enhanced activity‐dependent somatodendritic release of AVP, as shown by our biosensor sniffer cell studies. A shift in the balance of vasoactive signals during HF impacts the polarity of the salt‐evoked NVC Using both in vivo and ex vivo approaches, we previously demonstrated that the neurohormone AVP regulates vascular tone and reactivity of PAs in the SON by inducing vasoconstriction through its action on V1A receptors on vascular smooth muscle cells (Du et al., 2015). This included a vasoconstrictive NVC response to a systemic salt challenge (R. K. Roy et al., 2021). Notably, we also demonstrated that when AVP signalling was pharmacologically blocked, a previously masked vasodilatory response was revealed (R. K. Roy et al., 2021), which was likely mediated by NO (Du et al., 2015). This suggests that a salt challenge activates a mix of vasodilatory and vasoconstrictor signals, and that under normal conditions, the AVP‐mediated signal predominates, resulting in an overall vasoconstrictive response. Nonetheless, the masked vasodilatation may function to limit the duration and magnitude of this vasoconstrictive response, preventing excessive vasoconstriction during the sustained salt challenge, and ensuring dynamic vascular homeostasis. A significant finding of the present study is a drastic change in the polarity of the salt‐evoked NVC response in rats with HF. Rather than vasoconstriction being the dominant response, we found that the same homeostatic challenge (i.e. salt) resulted in a vasodilatory response of PAs in the SON. This vasodilatation was primarily mediated by adenosine acting on A2A receptors, which are known to be preferentially expressed on the vasculature (Dixon et al., 1996) and to promote vascular smooth muscle relaxation (Ponnoth et al., 2009). Interestingly, the adenosine‐induced vasodilatation in HF rats masks an underlying AVP‐mediated vasoconstriction, which became evident only when A2A receptor signalling was topically blocked within the SON. Thus, similar to control conditions, the salt challenge in HF rats resulted in the release of competing vasoactive signals. However, in this condition, vasodilatation predominated over vasoconstriction. Despite the predominant adenosine/A2A‐mediated vasodilatation, we found that the magnitude of the masked AVP‐mediated vasoconstriction was larger in HF rats compared with sham rats, suggesting an enhanced activity‐dependent somatodendritic release of AVP in the former. We and others have previously shown that the excitability of AVP neurons is enhanced in HF rats (Patel et al., 1993; Potapenko et al., 2011; Vahid‐Ansari & Leenen, 1998), which leads to increased circulating levels of the hormone (Iovino et al., 2018; Price et al., 2004). However, whether an augmented systemic release is accompanied by central somatodendritic release of AVP has not been tested before. Our studies using a biosensor sniffer cell support the notion that activity‐dependent central release of AVP is also enhanced in HF, a mechanism likely contributing to the more pronounced but ‘masked’ AVP‐mediated vasoconstriction in this condition. Still, whether this is directly due to the increased excitability of AVP neurons or other factors influencing somatodendritic release (e.g. intracellular Ca2+ homeostasis, increased AVP dendritic cargo) remains to be determined. An important consideration in interpreting the polarity shift of the salt‐evoked vascular response in HF rats is the potential contribution of intrinsic microvascular autoregulatory mechanisms. Increases in intravascular volume or perfusion pressure during systemic salt loading could, in principle, engage pressure‐ or stretch‐sensitive mechanisms within the microvasculature or astrocytic endfeet, leading to activity‐independent vasomotor responses (Attwell et al., 2010; Filosa & Iddings, 2013). However, several observations argue against intrinsic autoregulation as the primary driver of the responses observed here. First, the same salt challenge evoked opposite vascular responses in sham and HF rats, despite comparable systemic pressor responses, which would be expected to engage autoregulatory mechanisms in a similar direction across groups. Second, pharmacological blockade of adenosine A2A receptors unmasked a robust AVP‐dependent vasoconstriction in HF rats, indicating that the vasodilatory response reflects an active, receptor‐mediated process rather than a passive myogenic adjustment. Third, local blockade of AVP V1A receptors abolished both the adenosine‐mediated vasodilatation and the underlying AVP‐dependent vasoconstriction, demonstrating that intact AVP signalling is required for the vascular response to salt loading. Together, these findings support a model in which salt‐evoked changes in SON microvascular tone are dominated by activity‐dependent neuromodulatory signalling rather than intrinsic autoregulatory reflexes, although interactions between these mechanisms cannot be excluded. Microglia as a putative source of adenosine mediating salt‐evoked vasodilatation in HF rats There is accumulating evidence that adenosine and its receptors, particularly the A2 subtypes (A2A and A2B), contribute to CBF regulation (Kusano et al., 2010; Soricelli et al., 1995; Van Wylen et al., 2024), including NVC responses (Fu et al., 2025; Shi et al., 2005). For example, adenosine release in response to hypoxia leads to cortical pial vessel dilatation and increased cortical CBF, effects prevented by A2A receptor blockade (O'Regan, 2005). Moreover, a reduced CBF was observed in A2A receptor knock‐out mice (Ledent et al., 1997). Finally, adenosine receptors were also implicated in mediating NVC response in the barrel cortex in response to whisker stimulation (Shi et al., 2008). Together, these findings underscore the crucial role of adenosine, particularly through A2A receptor signalling, in regulating CBF and NVC responses. Adenosine is synthesized and released through multiple pathways, primarily from the breakdown of ATP (Fredholm, 2007). Microglia, the resident immune cells of the CNS, have recently emerged as a significant source of extracellular adenosine. Upon activation by various stimuli such as inflammation, hypoxia or neuronal activity, microglia release ATP, which is rapidly converted to adenosine by the ectonucleotidases CD39 and CD73 expressed on microglial surface (Matyash et al., 2017; Ransohoff, 2016). Notably, recent studies support the role of microglia in mediating NVC vasodilatory responses (Császár et al., 2022a; Ding et al., 2021; Fu et al., 2025). In the present study, we demonstrate that, in addition to A2A receptor blockade, the salt‐evoked vasodilatation in HF rats was also prevented by selective blockade of P2Y12 receptors. These are G‐protein‐coupled receptors, which are predominantly expressed on the surface of microglia in the CNS (Haynes et al., 2006 ; Mildner et al., 2017). P2Y12 receptors are essential for microglial functioning and their response to a variety of signals, including those from neuronal origins (Cserép et al., 2020; Davalos et al., 2005 ; Gu et al., 2016 ; Haynes et al., 2006 ; Sipe et al., 2016). Moreover, a recent study has shown that microglia‐mediated NVC responses in the cortex are dependent on the normal function of their P2Y12 receptors (Császár et al., 2022a). Based on these observations, we propose that microglia may present one potential source of adenosine contributing to the salt‐evoked vasodilatory responses in HF rats. The present study does not determine whether the P2Y12‐sensitive mechanism implicated here is specific to the SON or is shared by other hypothalamic or extra‐hypothalamic regions during HF. Thus, whether similar microglial or P2Y12‐sensitive mechanisms operate in adjacent hypothalamic or extra‐hypothalamic regions remains an important question for future studies. A contribution of microglia to the salt‐evoked vasodilatory response is reasonable in the context of HF. In this sense, we and others demonstrated that neuroinflammation, including increased microglial activation, constitutes a major pathophysiological feature in the hypothalamus and other brain regions during HF (Althammer et al., 2023; Badoer, 2010; Kang et al., 2008; Najjar et al., 2019; Rana et al., 2010). Moreover, we recently reported an increased proportion of vessel‐associated microglia in several brain regions of HF rats (Althammer et al., 2024b). Finally, in conditions of hypoperfusion/hypoxia, vessel‐associated microglia increased their process motility and contributed to adaptive responses to cortical CBF reduction, actions that were also shown to be dependent on P2Y12 receptors (Császár et al., 2022a). This is mechanistically relevant given that in the present study, we report a condition of sustained, basal hypoperfusion/hypoxia in the SON of HF rats. Together, these results suggest that the structural remodelling in the microglia/microvascular space during HF, along with the functional priming induced by the hypoxic milieu, could act as factors that facilitate the ability of microglia in this condition to influence vascular dynamics and to dominate the polarity of the NVC response. Still, while our results strongly support microglia as a source of adenosine mediating the salt‐induced vasodilatation observed in HF, we can't rule out other possible cellular sources, such as endothelial and/or neuronal ones. Salt‐evoked adenosine‐mediated vasodilatation in HF is functionally dependent on intact AVP/V1AR signalling Our studies indicate that an acute salt challenge in HF rats recruits opposing vasoactive influences within the SON: an AVP/V1AR‐dependent vasoconstrictive component and an adenosine/A2AR‐mediated vasodilatory component. However, whether these signalling mechanisms are activated in parallel by salt, or instead operate in a serial or permissive manner has not been fully established. One possibility is that salt independently stimulates both AVP neuronal activity and adenosine release from non‐neuronal elements. Indeed, high salt has been shown to promote microglial activation and neuroinflammatory responses in several contexts, including multiple sclerosis (Akaishi et al., 2025), ischaemia (Zhang et al., 2020) and stress‐related paradigms (Gilman et al., 2019). Alternatively, adenosine signalling may depend on, or be facilitated by, salt‐evoked AVP/V1AR signalling. Microglia express receptors for multiple neurotransmitters and neuropeptides (Pocock & Kettenmann, 2007) and are capable of sensing local neuronal activity. Consistent with this, oxytocin has been shown to modulate microglial function (Gonzalez & Hammock, 2022; Inoue et al., 2019 ; Mairesse et al., 2019 ; Panaro et al., 2020) and AVP has been reported to evoke Ca2 + responses in isolated microglial cells, particularly under pro‐inflammatory conditions (Pannell et al., 2014). These observations raise the possibility that salt‐evoked AVP release within the SON could indirectly promote adenosine signalling through glial mechanisms. Importantly, local blockade of AVP V1A receptors within the SON abolished all vascular responses to salt loading in HF rats, with neither vasodilatation nor vasoconstriction observed. This finding argues against a model in which the salt‐evoked adenosine/A2AR‐mediated vasodilatation is activated completely independently of AVP/V1AR signalling. If the adenosine‐mediated vasodilatory pathway were fully independent, V1AR blockade would be expected to preserve, or potentially enhance, the vasodilatory response by removing the opposing AVP‐dependent vasoconstrictive component. Instead, V1AR blockade eliminated the entire vascular response to salt. Thus, while these results do not establish a direct biochemical interaction between AVP and adenosine pathways, they support a functional dependence of the adenosine‐mediated vasodilatory response on intact AVP/V1AR signalling. The specific cellular intermediary responsible for adenosine production cannot be determined from the present data. Thus, although microglia are a plausible candidate, contributions from other cellular sources, including endothelial cells or neurons, cannot be excluded. Study limitations Several limitations of the present study should be acknowledged. First, a limitation of the present study is that the optical PO2 sensor provides a local estimate of oxygenation within the SON parenchyma but does not resolve the precise oxygen microenvironment surrounding the specific PA analysed for diameter changes. The probe was positioned within the SON using the ICA/MCA bifurcation and previously validated anatomical and vascular landmarks in this transpharyngeal preparation, providing confidence that the recordings were obtained from within the SON region. Therefore, the PO2 measurements should be interpreted as reflecting SON tissue oxygenation rather than vessel‐specific perivascular oxygen dynamics. Importantly, the direction of the PO2 changes was fully consistent with the vascular responses observed in each condition, with decreased PO2 accompanying vasoconstriction in sham rats and increased PO2 accompanying vasodilatation in HF rats, supporting the physiological relevance of these local SON measurements. A related limitation is that we did not perform parallel PO2 measurements in adjacent hypothalamic or extra‐hypothalamic regions. Thus, while the present data demonstrate that salt loading restores local oxygenation within the SON of HF rats, they do not establish whether this response is unique to the SON. Future studies using region‐specific PO2 measurements will be required to determine the anatomical specificity of this response. Second, although our pharmacological data support a role for adenosine signalling in shaping the polarity of the salt‐evoked vascular response in HF, the precise cellular source of adenosine within the SON cannot be definitively established. While microglia are a plausible contributor based on P2Y12 receptor involvement and prior literature, other potential sources, including endothelial cells and neurons, may also participate in adenosine production during salt loading. Third, the present experiments do not identify the specific targets through which adenosine and AVP exert their effects. Adenosine may act directly on vascular smooth muscle cells, indirectly via astrocytic signalling, or through capillary‐initiated mechanisms involving pericytes and/or conducted vascular responses. Dissecting these possibilities will require higher‐resolution imaging approaches and cell‐specific manipulations beyond the scope of the current study. Finally, all experiments were performed under urethane anaesthesia, which preserves key physiological reflexes and has been the anaesthetic of choice for in vivo studies of SON neuronal activity and function employing a transpharyngeal approach (Abbasi et al., 2022; Ludwig & Leng, 1997; Perkinson et al., 2022; Sabatier et al., 2004), including our initial characterization of salt‐evoked neurovascular responses using the same experimental preparation (R. K. Roy et al., 2021). Nevertheless, anaesthesia may influence baseline neurovascular tone, and vascular responses under awake conditions may differ in magnitude or kinetics. Future studies in awake or minimally restrained preparations will therefore be necessary to determine how behavioural state influences salt‐evoked neurovascular coupling in the SON. Despite these limitations, the present study identifies a state‐dependent reorganization of neurovascular signalling in the hypothalamic SON during HF, revealing how competing neuromodulatory pathways shape the polarity of salt‐evoked vascular responses. In summary, our results demonstrate that HF is associated with a basal hypoperfused and hypoxic state of the hypothalamic SON and a marked reorganization of salt‐evoked NVC. Under these conditions, the NVC response switches polarity, such that salt loading elicits a predominant vasodilatory response that partially restores local oxygen availability. This response is mediated, at least in part, by adenosine signalling via A2A receptors, with microglia as a plausible source. Consistent with recent studies in other brain regions (Császár et al., 2022b), these findings support the concept that neuroimmune–vascular interactions dynamically shape CBF and NVC, particularly under conditions of impaired perfusion. Given that compromised CBF is a common feature of cardiovascular and cerebrovascular diseases, including HF, stroke, hypertension and vascular dementia (Iadecola, 2017; Kisler et al., 2017; Wolters et al., 2017), the mechanisms identified here may have broader relevance for understanding neurovascular dysfunction across disease states. Additional information Competing interests None declared. Author contributions R.K.R.: conception and design of the work, acquisition, analysis or interpretation of data for the work, drafting the work and revising it critically for important intellectual content. E.C.L.: acquisition, analysis or interpretation of data for the work, drafting the work and revising it critically for important intellectual content. M.B.O.: acquisition, analysis or interpretation of data for the work. J.A.F.: conception and design of the work. J.E.S.: conception and design of the work, analysis or interpretation of data for the work, drafting the work and revising it critically for important intellectual content. Funding Grant support was provided by NIH R01HL162575 to J.E.S. and J.A.F.; American Heart Association 916907 to R.K.R. We also acknowledge support provided by the Centre for Neuroinflammation and Cardiometabolic Diseases at GSU. Generative AI statement OpenAI ChatGPT (GPT‐5.5 Thinking; OpenAI, San Francisco, CA, USA) was used only for limited language editing and clarity improvement of author‐generated text. The tool was not used to generate original scientific content, analyse data, create or modify figures, generate references or draw scientific conclusions. All AI‐assisted edits were reviewed and approved by the authors, who take full responsibility for the accuracy and integrity of the manuscript. Translational perspective Heart failure is associated with excessive neurohumoral activation, including increased release of the hormone vasopressin, which contributes to fluid retention and disease progression. We tested the hypothesis that heart failure alters neurovascular coupling in the supraoptic nucleus, a hypothalamic brain region that regulates vasopressin secretion and body fluid balance. Neurovascular coupling is the process by which local blood vessels adjust their diameter in response to neural activity. Contrary to our initial prediction that salt stimulation would worsen vasoconstriction and tissue hypoxia in heart failure, we found that the vascular response switched polarity: salt loading instead evoked vasodilatation and improved local tissue oxygenation. This adaptive response was mediated by adenosine acting through A2A receptors and depended functionally on intact vasopressin signalling, revealing a disease‐associated reorganization of vasoactive mechanisms within the hypothalamus. These findings suggest that neurovascular changes in brain circuits controlling fluid balance may influence the intensity of neurohumoral activation in heart failure and could be an underexplored contributor to disease progression or compensation. Defining whether this vasodilatory adaptation restrains vasopressin output, and whether its failure worsens fluid overload, could identify new central mechanisms to target alongside existing cardiac and renal therapies. Future studies should determine how hypothalamic neurovascular remodelling affects vasopressin release, establish whether similar mechanisms operate in patients or other autonomic brain regions, and test whether adenosine‐ or neurovascular‐based strategies can be leveraged to blunt maladaptive neurohumoral drive in heart failure. Supporting information Peer Review History