Structural and functional connectivity from the dorsomedial hypothalamus to the ventral medulla as a chronological amplifier of sympathetic outflow

Psychological stress activates the hypothalamus, augments the sympathetic nervous output, and elevates blood pressure via excitation of the ventral medullary cardiovascular regions. However, anatomical and functional connectivity from the hypothalamus to the ventral medullary cardiovascular regions has not been fully elucidated. We investigated this issue by tract-tracing and functional imaging in rats. Retrograde tracing revealed the rostral ventrolateral medulla was innervated by neurons in the ipsilateral dorsomedial hypothalamus (DMH). Anterograde tracing showed DMH neurons projected to the ventral medullary cardiovascular regions with axon terminals in contiguity with tyrosine hydroxylase-immunoreactive neurons. By voltage-sensitive dye imaging, dynamics of ventral medullary activation evoked by electrical stimulation of the DMH were analyzed in the diencephalon-lower brainstem-spinal cord preparation of rats. Although the activation of the ventral medulla induced by single pulse stimulation of the DMH was brief, tetanic stimulation caused activation of the DMH sustained into the post-stimulus phase, resulting in delayed recovery. We suggest that prolonged excitation of the DMH, which is triggered by tetanic electrical stimulation and could also be triggered by psychological stress in a real life, induces further prolonged excitation of the medullary cardiovascular networks, and could contribute to the pathological elevation of blood pressure. The connectivity from the DMH to the medullary cardiovascular networks serves as a chronological amplifier of stress-induced sympathetic excitation. This notion will be the anatomical and pathophysiological basis to understand the mechanisms of stress-induced sustained augmentation of sympathetic activity.

Anterograde tract-tracing. Anterograde tracing was performed in male Wistar rats (8-10-week old, n = 5) anesthetized with a mix of anesthetic agents (0.3 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol, i.p.). The tracer, biotinylated dextran amine (BDA) (Molecular Probes, Eugene, OR) (10% dissolved in 0.01 mol/L PB) was stereotaxically injected into the DMH as a single iontophoretic injection (5 µA, 400 ms, 1 Hz, 30 min) 11 . For the injection of BDA a pipette tip was placed at the site with its coordinates 2.6 mm caudal to the bregma, 0.4 mm right of the midline and 8.6 mm deep from the dorsal surface of the brain. After 1-week survival, the rats were deeply reanesthetized and perfused. Then, the brains were removed, postfixed, saturated with a 20% sucrose solution, and cut on a freezing microtome as described above.
One of every two series of sections was incubated in PBS containing 0.3% Triton X-100 and ABC-Elite for 1 h, followed by incubation in the solution consisting of 1.25 µmol/L biotinylated tyramide, 3 µg/mL of glucose oxidase, 2 mg/mL of beta-D-glucose, and 1% bovine serum albumin in 0.1 mol/L PB to intensify the BDA signals. Subsequently, sections were incubated in ABC-Elite for 1 h and then treated with ImmPACT SG solution (Vector Labs). BDA-labeled axons were stained dark blue to black. Finally, the sections were mounted, coverslipped, examined under a light microscope and photographed with a virtual slide scanner (PreciPoint, M8). BDA-labeled axons and terminals were plotted on photographs using a camera lucida.
Combination of anterograde tracing with immunohistochemistry for tyrosine hydroxylase. The anatomical relationship of DMH axon fibers and tyrosine hydroxylase (TH)-immunoreactive neurons in the medulla was investigated in another series of sections from 2 rats. When the BDA-labeled axons were visualized as described above, the sections were incubated overnight in avidin blocking solution (Avidin/Biotin Blocking Kit; Vector Labs) and then incubated in mouse monoclonal antibody for TH (1:2000; Immunostar, Hudson, WI) in biotin solution (Avidin/Biotin Blocking Kit) for 24 h. Subsequently, sections were incubated in biotinylated donkey anti-mouse IgG, incubated in ABC-Elite, and then reacted with DAB solution. They were mounted, coverslipped, and photographed using the light microscope.
In vitro preparation for voltage imaging. We used a diencephalon-lower brainstem-spinal cord preparation obtained from neonatal Wistar rats (either male or female, 0-2-day old, n = 9). The principal procedure for preparing the preparation has been previously described 12,13 . In brief, under deep isoflurane anesthesia, the brain and spinal cord were quickly isolated. The temporal lobes and other cerebral structures were removed. Then, the structures rostral to the level between the hypothalamic infundibulum and the bilateral basal regions of the temporal lobe were dissected (Fig. 1A). The dissection took place in the artificial cerebrospinal fluid (aCSF) consisting of (mmol/L) NaCl 124, KCl 5, KH 2 PO 4 1.2, CaCl 2 2.4, MgCl 2 1.3, NaHCO 3 26 , and glucose 30, equilibrated with 95% O 2 and 5% CO 2 (pH 7.4). For the staining with a voltage-sensitive dye, the preparation was incubated in aCSF containing a voltage-sensitive dye di-2-ANEPEQ (0.15-0.2 mmol/L) (Invitrogen, Carlsbad, CA) for 50 min while bubbling 95% O 2 and 5% CO 2 gas mixture [14][15][16] . Dissection of the preparation and its staining with the voltage-sensitive dye was conducted in aCSF at room temperature of 24-25 °C. www.nature.com/scientificreports/ Electrical stimulation and optical recording. After staining, the diencephalon-brainstem-spinal cord preparation was transferred into a recording chamber, mounted under an epifluorescence microscope (Optiphoto-2-UD, Nikon, Tokyo, Japan) with a 4 × objective lens (Plan Apo, NA 0.20, Nikon). Then, the preparation was fixed on the chamber floor, the ventral side up, with miniature pins, and was superfused with oxygenated aCSF of 26-28 °C at a rate of 4 mL/min 13,17 . The preparation's viability was confirmed by monitoring neural respiratory output from the C4 ventral root of the cervical spinal cord ( Supplementary Fig. 1). To stimulate the hypothalamus, a monopolar tungsten microelectrode (catalog# 573,400, A-M Systems, Carlsborg, WA) was stereotaxically inserted from the transected surface of the diencephalon, so that its tip was within the DMH region, according to a neonatal rat brain atlas 18 . The insertion point was set 0.3-0.5 mm lateral to the midline (third ventricle) and 0.1 mm deep from the surface (Fig. 1B). The microelectrode position was periodically checked, and adjusted when necessary, so that the electrically stimulated region coincided with the DMH, by confirming the concordance between the directly stimulated and depolarized area monitored by voltage imaging. Electrical stimulation was performed with two different modalities. One was a single pulse of 0.5-1 mA intensity and 3-5 ms duration, which were conducted in 5 preparations. Among these 5 preparations, four were used only in single pulse stimulation experiments. The stimulation intensity and duration were slightly adjusted depending on the responsiveness in each preparation. The other was tetanic stimulation of 0.4 mA intensity, 500 μs pulse duration, 100 Hz frequency, and 10 s tetanic stimulation duration, which was conducted also in 5 preparations. Among these 5 preparations four were used only in tetanic stimulation experiments, and one for both single pulse and tetanic stimulation experiments.
To detect stimulation-induced neural activation, fluorescence signal in the ventral medulla was continuously captured by optical recording system. The imaged ventral medullary area is indicated with a circle in Fig. 1B. The preparation was illuminated with a tungsten-halogen lamp (150 W) through a band-pass excitation filter (λ = 480-550 nm), and epifluorescence was detected through a long-pass barrier filter (λ > 590 nm) with a CMOS sensor array (100 × 100 μm pixel size, 100 × 100 pixel array; MiCAM Ultima, BrainVision, Tokyo, Japan) 11 . The optical signal was sampled at a frame rate of 50 Hz in single pulse stimulation and 12.5-25 Hz in tetanic stimulation. In the former mode of stimulation, optical signal was recorded for 5.12 s, starting at 1.28 s before the www.nature.com/scientificreports/ onset of stimulation. In the latter, the optical signal was recorded for 40.96 s, starting at 10.24 s before the onset of stimulation.
Statistical analysis. Imaging data were analyzed using an image analysis software, BV Analyzer (BrainVision). Details of imaging data analyses were as we described previously 11,[14][15][16]19 . A change in fluorescence intensity (ΔF) relative to initial intensity (F0) in each pixel was calculated. To normalize the difference in the amount of membrane-bound dye and illumination within the preparation, background fluorescence intensity at each pixel was divided by the maximum background fluorescence. Then, ratio of ΔF to the normalized background fluorescence intensity (F) was calculated at each pixel in each frame (ΔF/F). If F was less than 0.25, ΔF⁄F was set to zero. A negative ΔF⁄F corresponds to membrane depolarization. To increase the signal-to-noise ratio, cycle triggered averaging was conducted for 20-50 cycles with 7-20 s intervals in the single pulse stimulation and 5 cycles with 1.5-5 min intervals in tetanic stimulation. To evaluate the depolarizing neural activity in the ventral medulla evoked by hypothalamic stimulation, changes in epifluorescence were quantified at four squared areas in Fig. 1B (each 0.5 mm × 0.5 mm), i.e., ipsilateral RVLM, ipsilateral caudal ventrolateral medulla (CVLM), ipsilateral rostral ventromedial medulla (RVMM), and the raphe pallidus (RP). We determined these regions at the level of the XIIth cranial nerve root-1.3 mm lateral to the midline, the XIIth cranial nerve root-0.3 mm lateral to its root, the same level of the RVLM-0.5-1.0 mm lateral to the midline, and in the midline area at the level between the Xth and XIIth cranial nerve roots, respectively. Stimulation-induced excitation was defined as a median value of ΔF/F twice larger than the baseline value. The duration of excitation was used as an index of post-stimulus 75% recovery time, which was defined between the endpoint of stimulation and the point of a median ΔF/F decay by 75% off its peak. Data were presented as means ± SE. Differences between poststimulus 75% recovery time in the single pulse and tetanic stimulation were analyzed with the Welch t-test. A p value < 0.05 defined statistically significant differences.

Ethics approval.
All experiments were carried out in accordance with the National Institutes of Health

Results
Histology. We first examined the distribution of RVLM-projecting neurons in the DMH, using a retrograde tracing technique by injection of FG into the RVLM. The accuracy of the injection site was confirmed by histologically observing that the FG deposit was limited in the RVLM ( Fig. 2A). Retrogradely FG-labeled neurons were observed bilaterally in the hypothalamus with a clear-cut ipsilateral dominance. At the level of the DMH, these neurons were distributed sparsely in the tuberal hypothalamus dorsal and lateral to the fornix, in the peduncular part of the lateral hypothalamus, and in the posterior hypothalamic nucleus (Fig. 2, Supplementary  Fig. 2). We further examined the neuronal projection from the DMH to the medulla oblongata by anterograde tracing. In two rats that were most optimally injected with BDA into the DMH, in which the BDA deposit was confined to the DMH (Fig. 3A, Supplementary Fig. 3), anterogradely BDA-labeled fibers and axon terminals were distributed bilaterally, also with a clear-cut ipsilateral dominance, in the reticular formation of the medulla oblongata. In addition, a few labeled axons were seen in the rostrocaudal extent of the nucleus of the solitary tract ( Fig. 3B-D). Many labeled fibers were found in the ventromedial medulla, which extended to the mid-medullary region. Also, a few labeled axons were present in the ventrolateral medulla, on the surface of the pyramidal tract, and at the boundary between the pyramidal tract and ventromedial medulla (parapyramidal region) (Fig. 3B-E). In the RVLM, there is a cluster of tyrosine hydroxylase (TH)-immunoreactive catecholaminergic neurons. We found bouton-like varicosities labeled with BDA were in close apposition to the dendrites of TH-immunoreactive neurons ( Fig. 3F-I, Supplementary Fig. 4).

Voltage imaging.
Depolarizing neural activities on the ventral medulla induced by single pulse and tetanic electrical stimulation of the DMH were visualized by voltage imaging. The accuracy of the stimulating microelectrode placement was histologically confirmed by observing that the damaged region corresponded to the DMH ( Supplementary Fig. 5). A representative image of the single pulse stimulation of the hypothalamus is shown in Fig. 4. This stimulation induced depolarizing optical signals on the RVLM and RVMM in all specimens, whereas activation of the CVLM was detected in three and that of the RP in four out of the five specimens, respectively.
In response to tetanic stimulation of the DMH, neural activation was detected in the RVLM, RVMM, and RP in all specimens, whereas activation of the CVLM was seen in four out of the five specimens (Fig. 5).
Of note, tetanic, but not single pulse, stimulation of the DMH induced a delayed recovery from the RVLM activation after cessation of stimulation. In tetanic stimulation, post-stimulus 75% recovery time was significantly longer, as compared with that in single pulse stimulation, in the RVLM, CVLM, RVMM, and RP areas  (Fig. 6).

Discussion
In this study, tract-tracing revealed that DMH neurons send axonal projection directly to the ventral medullary cardiovascular regions and that some DMH neurons terminate onto dendrites of TH-immunoreactive neurons in the RVLM. In voltage imaging, both single pulse and tetanic electrical stimulation of the DMH induced www.nature.com/scientificreports/ depolarizing neural activity in the ventral medullary cardiovascular regions. However, recovery of medullary activity after cessation of stimulation was remarkably delayed after tetanic, but not after single pulse stimulation, indicating that the connectivity from the dorsomedial hypothalamus to the medullary cardiovascular networks serves as a chronological amplifier of stress-induced sympathetic excitation. The ventral medullary regions to which DMH neurons sent axonal projections coincided with the regions depolarized by stimulation of the DMH in voltage imaging experiments.
The major finding of this study is that the DMH and several ventral medullary regions that are important in cardiovascular regulation (RVLM, CVLM, RVMM and RP) are anatomically and functionally connected. Further, we anatomically confirmed for the first time that DMH neurons directly innervate catecholaminergic neurons in the RVLM. Several other histological studies have also reported a monosynaptic connection from the DMH to the dorsal motor nucleus of the vagus, nucleus ambiguus, and parvocellular reticular formation of the medulla 20 . Lindberg et al. 21 have shown synaptic contacts between the ventromedial hypothalamic neuronal efferents and TH-positive catecholaminergic neurons in the RVLM and the nucleus of the solitary tracts. However, the connectivity between the DMH efferents and TH-immunopositive neurons in the RVLM or other cardiovascular regions in the medulla have not been reported. In acute stress, the DMH plays a vital role in integrating autonomic, neuroendocrine, and behavioral responses 5 . Lisa et al. 22 have demonstrated that inhibition of the DMH abolishes the stress-induced cardiovascular response. Sympathetic premotor neurons in the RVLM are excited by DMH disinhibition 7,23 . Also, medullary raphe neurons mediate sympatho-excitation evoked by DMH disinhibition 7,24 . While we did not measure blood pressure or heart rate, we, as well as the above mentioned previous studies, have demonstrated that hypothalamic stimulation excitates the ventral medullary areas that are www.nature.com/scientificreports/  www.nature.com/scientificreports/ involved in stress-induced cardiovascular responses. These findings help understand how the ventral medulla is activated by acute psychological stress. In this study, tetanic hypothalamic stimulation excited not only the RVLM, RVMM and RP but also the CVLM. Our tract-tracing experiments revealed the anatomical connectivity from the DMH to the CVLM. It is well established that arterial baroreceptor afferents are excited by blood pressure elevation. They project to the nucleus of the solitary tract and provide a major excitatory input to GABAergic CVLM neurons. These neurons, in turn, inhibit the activity of presympathetic RVLM neurons 25 . Thus, the CVLM has a sympatho-inhibitory role. However, it has also been reported that neurons in the caudal pressor area (CPA), which is located just caudal to the CVLM, are sympatho-excitatory 26,27 . CPA neurons mediate the pressor response by inhibiting RVLM-projecting inhibitory units and activating RVLM-projecting excitatory CVLM units 27 . Among the ventral medullary regions activated by DMH stimulation, which was detected by our voltage imaging, the most caudal area could correspond to the CPA, suggesting the possible excitatory innervation of the CPA by the DMH.
It is noteworthy that tetanic stimulation of the hypothalamus induced post-stimulus sustained excitation in the ventral medulla, plausibly related to blood pressure regulation. However, our observations do not address the exact mechanism of delayed post-stimulus recovery. In the medullary cardiovascular regions, there exist a high density of angiotensin II type 1 receptors 28,29 . Glass et al. 30 have reported that angiotensin II-induced blood pressure elevation could be a phenotype of neural plasticity. Recent studies have demonstrated that not only neurons but also glial cells, especially astrocytes, are critically involved in neural plasticity 19,31,32 . Neurotransmitters, released from presynaptic terminals, activate nearby astrocytes in a paracrine manner. The activated astrocytes augment neuronal synaptic transmission by releasing gliotransmitters 33 . These phenomena occur within a few seconds but could persist for more than several minutes. Thus, astrocytes may also be involved in the blood pressure elevation evoked by psychological stress and, particularly, in the post-stress sustained blood pressure elevation. Voltage imaging could detect both neuronal and astrocytic excitation 19,34 . However, in the present study we could not distinguish optical signals from neurons and astrocytes. Further studies are needed to clarify the distinctive roles of neurons and astrocytes in the hypothalamic activation-induced sustenance of ventral medullary excitation, using different experimental designs, for instance, by calcium imaging.
Acute psychological stress induces elevation of blood pressure, which persists after the stress is relieved 35 . Recently, several studies have reported that the time course of post-stress cardiovascular recovery is predictive of future hypertension and other cardiovascular events 1,36,37 . Thus, a study of post-stress cardiovascular recovery contributes to the elucidation of the pathophysiology of hypertension. Our finding that tetanic stimulation of the hypothalamus induces post-stimulus sustained activation of the ventral medullary cardiovascular regions indicates that the connectivity from the DMH to the medullary cardiovascular networks serves as a chronological www.nature.com/scientificreports/ amplifier of stress-induced sympathetic excitation, and gives insight into the mechanisms underlying the persistence of post-stress sympathoexcitation. Among the medullary regions that are anatomically and functionally connected with the DMH, the RVMM is also important in processing nociceptive information including both antinociception and pronociception 38,39 . Wagner et al. reported that the DMH mediates stress-induced hyperalgesia via the RVMM 39 . Stress-induced DMH excitation may contribute to pain-induced blood pressure elevation via the activation of medullary regions including the RVMM, and this issue must be further clarified.
There may be a criticism that the tetanic stimulation-induced depolarizing activity in the ventral medulla is simply non-specific post-tetanic facilitation. However, in our number of preliminary trials of stimulating the diencephalon at the DMH as well as at many other nearby sites other than the DMH, we did not obtain positive responses to stimulation of sites other than the DMH. Thus, the tetanic stimulation-induced depolarizing activity in the ventral medulla could not be non-specific post-tetanic facilitation.
The experimental method of the present study has some limitations. Voltage imaging in the isolated diencephalon-lower brainstem-spinal cord preparation is performed in the condition devoid of blood circulation, in which the tissue PO 2 and pH environment are maintained within the physiological range in the layer only up to several hundred micrometers from the surface 40 . However, the function of the sympathetic neural network of the ventral medulla and spinal cord is reported intact in the preparation 41 . Further, it has been confirmed that the respiration-facilitating function of the diencephalon also remains intact in this preparation 13,42 . To further improve hypothalamic tissue oxygenation, we removed the brain structures rostral to the point of stimulation (Fig. 1). In this study, we performed electrical stimulation of the DMH in the neonatal rat preparation to mimic psychological stress. It has been reported that functional connectivity from the ventral medullary cardiovascular regions to the intermediate lateral cell column of the spinal cord is also intact in the neonatal rat brainstemspinal cord preparation 41,43 . Therefore, it is conceivable that the use of our in vitro preparation is suitable for the evaluation of functional connectivity between the hypothalamic and ventral medullary cardiovascular regions.
Although we obtained interesting findings by recording depolarizing activities in the cardiovascular regions in the ventral medulla, neither sympathetic nerve activity nor blood pressure was recorded in this reduced www.nature.com/scientificreports/ preparation, which is a major limitation when interpreting these results. The functional experimental findings obtained in the present study should be confirmed in in vivo experiments by monitoring sympathetic nerve output and blood pressure. Delayed recovery of ventral medullary excitation after cessation of tetanic stimulation of the hypothalamus, as observed in the present study, may have to do with the post-stress persistent sympathetic activation and the development of hypertension. In the present study, we stimulated the hypothalamus only for 10 s. Chronic psychological stress usually acts over a longer time scale of months or even years, resulting in hypertension. Nonetheless, Hall 44 has indicated that stress affects blood pressure also in a matter of seconds and the accumulation of short sympathetic responses underlies the development of hypertension. The mechanism of post-stimulus sustained excitation in the ventral medulla should be investigated over different time scales to better understand the pathophysiology of stress-induced hypertension.
Elucidation of the post-stress autonomic neural responses could provide a clue for the pathophysiologic mechanism of stress-induced hypertension. In this study, we demonstrated that neurons in the DMH send axonal projections to the ventral medullary cardiovascular regions. Besides, electrical stimulation of the DMH induces neural excitation in the ventral medullary cardiovascular regions and recovery from the excitation is delayed after tetanic stimulation is applied. We identified ventral medullary areas that are anatomically and functionally connected with the DMH. Thus, the connectivity from the DMH to the medullary cardiovascular networks serves as a chronological amplifier of stress-induced sympathetic excitation. To further understand the hypothalamus-ventral medullary connectomics, cells and transmitters involved should be investigated, with attention directed toward astrocytes and angiotensin II.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.