Switching control of sympathetic activity from forebrain to hindbrain in chronic dehydration

Non-technical summary

Dehydration, a life-threatening condition, occurs when the body does not replace adequate water lost through urination, sweating or when ill with diarrhoea. This presents the body with a major challenge of maintaining blood pressure – essential for consciousness that is dependent on the degree of body hydration, which dictates blood volume. We know that a major control mechanism involves a brain region called the hypothalamus that automatically maintains blood pressure. Our study has described the gene networks in key brain regions involved in the response to dehydration. We reveal a new structure in the brain that regulates blood pressure in dehydration and a unique genetic mechanism that exists within it. Moreover, our study unearths a remarkable form of flexibility within the brain during dehydration that involves switching control of blood pressure between two spatially distinct structures. We have provided new mechanistic insight to explain how the brain maintains body stability in face of the significant challenge of low water content.

Abstract

Abstract

We investigated the mechanisms responsible for increased blood pressure and sympathetic nerve activity (SNA) caused by 2–3 days dehydration (DH) both in vivo and in situ preparations. In euhydrated (EH) rats, systemic application of the AT₁ receptor antagonist Losartan and subsequent pre-collicular transection (to remove the hypothalamus) significantly reduced thoracic (t)SNA. In contrast, in DH rats, Losartan, followed by pre-collicular and pontine transections, failed to reduce tSNA, whereas transection at the medulla–spinal cord junction massively reduced tSNA. In DH but not EH rats, selective inhibition of the commissural nucleus tractus solitarii (cNTS) significantly reduced tSNA. Comparable data were obtained in both in situ and in vivo (anaesthetized/conscious) rats and suggest that following chronic dehydration, the control of tSNA transfers from supra-brainstem structures (e.g. hypothalamus) to the medulla oblongata, particularly the cNTS. As microarray analysis revealed up-regulation of AP1 transcription factor JunD in the dehydrated cNTS, we tested the hypothesis that AP1 transcription factor activity is responsible for dehydration-induced functional plasticity. When AP1 activity was blocked in the cNTS using a viral vector expressing a dominant negative FosB, cNTS inactivation was ineffective. However, tSNA was decreased after pre-collicular transection, a response similar to that seen in EH rats. Thus, the dehydration-induced switch in control of tSNA from hypothalamus to cNTS seems to be mediated via activation of AP1 transcription factors in the cNTS. If AP1 activity is blocked in the cNTS during dehydration, sympathetic activity control reverts back to forebrain regions. This unique reciprocating neural structure-switching plasticity between brain centres emphasizes the multiple mechanisms available for the adaptive response to dehydration.

Introduction

Blood volume and plasma osmolality are controlled tightly and robustly defended in mammals by homeostatic processes that integrate the responses of the cardiovascular, renal and neuroendocrine systems (Antunes-Rodrigues et al. 2004; McKinley et al. 2004). Regulation of osmotic stability and blood volume and hence arterial pressure are mediated by actions of both the antidiuretic hormone – vasopressin (VP; Burbach et al. 2001; Antunes-Rodrigues et al. 2004) and the sympathetic nervous system (e.g. Coote, 2005; Antunes et al. 2006; Osborn et al. 2007; Stocker et al. 2008; Toney & Stocker, 2010). Whilst the suproptic nucleus of the hypothalamus controls vasopressin release, the parvocellular neurons of the paraventricular nucleus (PVN) regulate sympathetic nerve activity (SNA) and consequently blood volume and arterial pressure (Sawchenko & Swanson, 1982; Yang & Coote, 1999; Hardy, 2001; Coote, 2005) via projections to the rostral ventrolateral medulla and/or intermediolateral cell column (Sawchenko & Swanson, 1982).

The PVN plays a role in controlling sympathetic nerve activity in both physiological situations, such as responses to hyperosmotic stimuli applied systemically (Antunes et al. 2006; Shi et al. 2008) and angiotensin II applied centrally (Latchford & Ferguson, 2003; Cato & Toney, 2005; Ferguson, 2009), and pathological conditions of hypertension (Allen, 2002) and heart failure (Patel, 2000; Guggilam et al. 2008; Badoer, 2010). Far less is known about the role of central AT₁ receptors and the PVN during dehydration. The current dogma is that there is an increased role for angiotensin II and the PVN in regulating sympathetic activity to support the hypertension associated with dehydration (Stocker et al. 2005, 2006, 2008; Freeman & Brooks, 2007; Osborn et al. 2007) since PVN inactivation caused a significant decrease in arterial blood pressure in dehydrated rats compared to controls (Stocker et al. 2005).

Our aim was to re-investigate the central neural control of sympathetic nerve activity and vasomotor tone after 2 or 3 days dehydration (DH) in unanaesthetized rats in vivo and in situ (Antunes et al. 2006; Colombari et al. 2010). We confirmed that in euhydrated (EH) rats, arterial pressure and SNA are, in part, dependent on the hypothalamus and AT₁ receptors. However, contrary to previous studies, the control of SNA in DH rats is AT₁ receptor independent and transfers predominantly to the commissural nucleus tractus solitarii (cNTS). To better understand the mechanism involved, a microarray analysis was performed on the NTS. One gene up-regulated by dehydration was the AP1 transcription factor JunD as described previously (Ji et al. 2007). We tested the hypothesis that AP1 activity was responsible for DH-induced functional plasticity in the cNTS. Following chronic blockade of its activity in the cNTS of DH rats, inactivation of the cNTS no longer had an effect on SNA; we show that it reverted back to the hypothalamus. Our results demonstrate a remarkable neural structural-switching plasticity within the brain that underpins the adaptive response to dehydration.

Methods

The authors have read, and the experiments comply with, the policies and regulations of The Journal of Physiology given by Drummond (2009).

Animals

All procedures conformed to the UK Animals (Scientific Procedures) Act 1986 and were approved by the University of Bristol Ethical Review Committee. Male rats (10–12 weeks old; Harlan Sera-lab, Loughborough, UK) were maintained in standardised temperature (22 ± 1°C), humidity (50 ± 5%) and diurnal conditions (10 h light, 14 h dark; lights on at 07.00). Control EH animals had access to both food and drinking water ad libitum. Dehydration involved complete fluid deprivation for 3 days (72 h; onset 11.00 am on day 1, end 11.00 am on day 3). EH animals were killed at the same time. In some in vivo studies DH was for 48 h. All DH rats had access to food.

Radiotelemetry recordings

Adult male Sprague–Dawley (250–270 g) rats were allowed to adapt for 1 week (standard laboratory rat chow with water ad libitum) before implantation of a radio-telemetry device for recording of arterial pressure (Data Sciences International, Arden Hills, MN, USA). Following procedures published previously (Waki et al. 2003; Antunes et al. 2006) rats were anaesthetized with ketamine (60 mg kg⁻¹) and medetomidine (250 g kg⁻¹) intramuscularly. A radio transmitter (Data Sciences International; TA11PAC40) was implanted to record arterial pressure chronically from the abdominal aorta. Anaesthesia was reversed with atipamezole (1 mg kg⁻¹). Animals were allowed to recover for 7 days prior to implementation of the 3 day dehydration protocol. Telemetry data were acquired and spectra analysed using Hey-Presto software (Waki et al. 2006) running on a PC laptop. Arterial blood pressure was recorded for a minute interval every hour for 24 h during the entire experimental period. Data are expressed as the mean ± standard error of mean (SEM) and were analysed by one-way ANOVA, followed by Dunnett's multiple comparison post hoc test for multiple comparisons.

Decorticate, un-anaesthetized, arterially perfused in situ preparation of rat

Male Wistar rats weighing 60–90 g were prepared and used to record activity of the thoracic sympathetic chain as described previously (Paton, 1996; Antunes et al. 2006; Colombari et al. 2010). Briefly, rats were heparinized (1000 units, given i.p.) and subsequently anaesthetized deeply with halothane until loss of their paw withdrawal reflex. They were decerebrated pre-collicularly and anaesthesia withdrawn. They were bisected sub-diaphragmatically and the head and thorax immersed in ice-chilled carbogenated Ringer solution. The phrenic nerve was cut distally. Preparations were transferred to a recording chamber. A double-lumen catheter (DLR-4, Braintree Scientific, MA, USA) was inserted into the descending aorta for retrograde perfusion. Perfusion was supplied via a peristaltic roller pump (Watson Marlow 505D) and consisted of carbogenated Ringer solution at 31°C (for constituents, see below). The second lumen of the catheter was used to monitor aortic perfusion pressure. The baseline perfusate flow was pre-set between 21–24 ml min⁻¹ and adjusted according to the absolute size of the animal. Euhydrated rats were perfused with isosmotic Ringer solution containing (mm): NaCl 120, NaHCO₃ 24, KCl 3, CaCl₂ 2.5, MgSO₄ 1.25, KH₂PO₄ 1.25, glucose 10; pH 7.3 after carbogenation. The osmolality, measured by a freezing point depression osmometer (Camlab, Roebling Micro-osmometer, Cambridge, UK), was 290 mosmol (kg water)⁻¹. Three day DH rats were perfused with a hyperosmotic ionic Ringer solution, prepared by adjusting the final concentration of NaCl resulting in an osmolality of 340 mosmol (kg water)⁻¹. This was identical to the observed plasma osmolality after 3 days dehydration in juvenile rats (340 ± 5 mosmol (kg water)⁻¹, n = 5). In some DH rats, we switched the perfusate from 340 to 290 mosmol (kg water)⁻¹; i.e. isotonic perfusate). Vasopressin (400–700 pm), to elevate perfusion pressure, was added in the perfusate in equivalent amounts for both groups (EH: 40.6 ± 5.5 μl and DH: 40.5 ± 3.3 μl). Thoracic (t; T8–T10 spinal level) and in some studies lumbar (L) sympathetic nerve activity (SNA) was recorded using bipolar glass suction electrodes together with the phrenic nerve. Signals were amplified (20 kHz, Neurolog) and filtered (50–1500 kHz, Neurolog). To standardize the data across preparations, noise levels were removed from the raw signal. Lidocaine (2%) was applied to the sympathetic chain at the end of the experiment so that the level of noise for each recording could be measured. The baroreceptor reflex was stimulated by transiently increasing perfusion rate to raise perfusion pressure and peripheral chemoreceptors stimulated using sodium cyanide (i.a. 50–100 μl 0.03%; Paton, 1996). Data were digitized using a CED 1401 A–D interface (CED, Cambridge Electronic Design, Cambridge, UK) and a computer running Spike 2 software (CED) with custom-written scripts for data acquisition and on- and off-line analyses. SNA changes were expressed as absolute levels and as percentage difference from baseline. Phrenic-triggered averaging (50 consecutive cycles) of integrated tSNA was carried out off-line (as Simms et al. 2009). The respiratory cycle was divided into inspiratory and expiratory periods to assess the level of tSNA during these phases. Neural inspiratory minute ventilation was taken as area under the curve of integrated phrenic discharge multiplied by frequency. All data were expressed as the mean ± standard error of mean (SEM). One-way or two-way (as appropriate) ANOVA followed by Student–Newman–Keuls post hoc or Student's t test were used for comparisons. Differences were taken as significant at P < 0.05.

Brain transection in situ

Three sequential transections were performed: (i) rostral to the superior colliculus (to remove the hypothalamus); (ii) caudal to the inferior colliculus to remove rostral pons; and (iii) medulla oblongata–spinal cord junction. Histological analysis was performed to assess the actual level of the transections (see below).

In vivo rats

For PVN studies, rats (280–320 g) were anaesthetized with ketamine (80 mg kg⁻¹) and xylazine (7 mg kg⁻¹, i.p.). In a stereotaxic frame the skull was levelled between Bregma and Lambda. Stainless-steel 23-gauge guide cannulae were implanted bilaterally into the PVN (1.8 mm caudal to Bregma, 0.5 mm lateral to midline and 5.3 mm below the dura) so that their tips were 2 mm above the PVN. After 5 days rats were re-anaesthetized with ketamine and xylazine and a femoral artery and vein cannulated. Water was removed from the cage of the DH rats. Two days later while recording arterial pressure (Powerlab 16SP, ADInstruments), muscimol (RBI, USA) dissolved in saline was injected bilaterally (100 nl of 100 pmol (100 nl)⁻¹) into the PVN of conscious rats using a 1 μl Hamilton syringe connected to an injector needle that was 2.0 mm longer than the guide cannula. For the in vivo cNTS study, rats (280–320 g) were first fitted with femoral artery and vein catheters (under ketamine and xylazine anaesthesia) and following 2 days dehydration were re-anaesthetized with urethane (830 mg kg⁻¹) and α-chloralose (55 mg kg⁻¹i.v.). (It was technically not possible to inject into cNTS in conscious dehydrated rats.) In a sterotaxic frame with the incisor bar 11 mm below the interaural line, the dorsal surface of the medulla was exposed and the cNTS microinjected with 25–30 nl of muscimol (100 pmol (100 nl)⁻¹) via a glass micropipette (0.5 mm caudal to the calamus scriptorium, midline and 0.5 mm ventral to the dorsal surface) while monitoring arterial pressure. Animals were maintained at 38°C using a thermostatically controlled heating blanket. For both PVN and cNTS, measurements were made 20 min after muscimol injections; sites were marked with Evans blue dye (1%, 100 nl or 30 nl).

Microinjection into cNTS in situ

The head of the preparation was fixed by ear bars and a nasal clamp so that the brainstem was horizontal. Microinjections (in situ: 20 mm isoguvacine, 60 nl) were made into midline cNTS 300–500 μm caudal to calamus scriptorius and 300–400 μm ventral to the dorsal surface of the brainstem using a glass micropipette. The volume microinjected was determined by viewing the movement of the meniscus through a binocular microscope fitted with a pre-calibrated eyepiece reticule.

Histology of in situ and in vivo studies

At the end of the experiments, 1–2% (w/v) Evans blue solution was injected into the PVN and cNTS using identical volumes to those of isoguvacine and muscimol. For these and the transection studies, the brainstem was removed and placed for 2–3 days in 10% buffered formalin. It was then frozen and cut coronally (50 μm sections). Sections were counter-stained with Giemsa and analysed by light microscopy to confirm injection sites and transection levels.

In vivo gene transfer into cNTS

Ad-CMV-DNFosB-IRES-eGFP (2.9 × 10⁶ pfu μl⁻¹) is a replication-deficient recombinant adenoviral vector expressing a myc-tagged FosB dominant negative (the kind gift of Yasuko Yamamura, Tokyo Medical and Dental University, Japan; Yamamura et al. 2000) under the control of the human cytomegalovirus (CMV) promoter. Enhanced green fluorescent protein (eGFP) is also expressed from this vector via an internal ribosome entry site (IRES). Ad-CMV-eGFP (1.6 × 10⁶ pfu μl⁻¹) was used as a control. In rats anaesthetized with ketamine (60 mg kg⁻¹) and medetomidine (250 μg kg⁻¹) intramuscularly, two 200 nl microinjections of viral suspension were made into the cNTS at two separate sites spanning 500 μm rostral/caudal to the calamus scriptorius (0.0 μm and 500 μm) and 400 μm below the dorsal surface. Each injection was made over 1 min. Anaesthesia was reversed with atipamezole (1 mg kg⁻¹) and animals were allowed to recover for 5–7 days before 3 days of dehydration. At the end of the third day of dehydration, animals were prepared for the in situ preparation as described above.

Immunohistochemical detection of eGFP or myc

The brain was removed and left 2 days in 4% (w/v) paraformaldehyde in 0.1 m PBS at 4°C, followed by another 2 days in the cryoprotected solution of 20% (w/v) sucrose in 4% (w/v) paraformaldehyde in 0.1 m PBS at 4°C. The following morning the brains were rapidly frozen over liquid nitrogen and four sets of coronal sections (30 μm) of the cNTS were sectioned on a cryostat (Cryocut CM1900, Leica, Switzerland). The free-floating sections were collected in 24-well tissue culture plates containing PBS. Antigens were detected using rabbit polyclonal GFP antiserum (Invitrogen, Renfrew, UK; catalogue no. A6455) or mouse monoclonal anti-myc (Invitrogen, catalogue no. R950–25). The anti-myc antibody permits detection of recombinant proteins containing the myc epitope (-Glu-Gln-Lys-Leu-Ile-Ser-Glu-Glu-Asp-Leu-). Free-floating rat brainstem sections were incubated for 15 min in a blocking solution comprising 10% (v/v) normal goat serum (NGS, Sigma, St Louis, MO, USA) and 0.3% (v/v) Triton X-100 (Sigma) in 0.1 m PBS followed by rinses in PBS (3 × 10 min). Sections were then incubated in a polyclonal rabbit anti-GFP primary antiserum (1:2000 dilution) or in a mouse monoclonal anti-myc (1:500 dilution) in PBS containing 1% (v/v) NGS and 0.3% (v/v) Triton X-100 for 24 h at 4°C. After the primary antibody incubation the sections were rinsed in PBS (3 × 10 min) prior to 1 h incubation in goat anti-rabbit AlexaFluor 488 (1:500; Molecular Probes, Eugene, OR, USA) or goat anti-mouse AlexaFluor 594 (1:500, Molecular Probes). Following further rinses in PBS (3 × 5 min) sections were mounted onto slides in 0.5% (w/v) gelatin and allowed to air-dry for 10–15 min before being coverslipped using an antifade fluorescent mountant (VectorShield, Vector Laboratories. Peterborough, UK). The sections were visualized on a fluorescent microscope with the appropriate filter.

Microarray analysis

Area postrema (AP) and NTS were carefully dissected from EH and DH rats, processed and hybridised to Rat Affymetrix Genechip 240 2.0 microarrays in collaboration with Source Bioscience Geneservice (Nottingham, UK) as previously described (Hindmarch et al. 2006). Each microarray was hybridized with RNA extracted from five animals and was repeated with five independent replicates resulting in five independent EH arrays and five independent DH arrays for each brain structure. Primary analysis of microarrays in GeneSpring GX11 (Agilent Technologies UK, Stockport, UK) revealed that one AP-DH array, one NTS-EH array and one NTS-DH array failed quality control and were removed from further analysis resulting in n = 4 for these groups. The raw .CEL files were imported into GeneSpring GX11 for normalisation using the MAS5 summarization algorithm and baseline transformation to the median of all samples. Data were filtered so that only probe-sets that are considered to be present in all the chips from at least one of the experimental groups were available for statistical analysis. A one-way (Welch) ANOVA was performed which assumed unequal variance and included the Benjimini & Hochberg multiple testing correction, limiting those genes identified by chance to just 5%. Post hoc Tukey HSD analysis revealed those genes that were significantly regulated by dehydration in either tissue.

Data deposition

Raw data have been submitted to the National Centre for Biotechnology Information Gene Express Omnibus database. http://www.nbci.nlm.nih.gov/geo (Data are accessible for download at: http://www.vasopressin.org/#/data-bank/3755442).

Results

Chronic dehydration increases arterial pressure and sympathetic nerve activity (SNA)

We used radio-telemetry to non-invasively measure pulsatile arterial pressure during DH (Fig. 1A). Superimposed on a circadian rhythm, an upward trend in arterial pressure is evident with a significant increase in mean arterial pressure (MAP) observed from the second day of DH (114 ± 1.5 and 116 ± 1.4 vs. baseline 102 ± 1.1 mmHg, respectively, 2 and 3 days DH; n = 8, P < 0.05, Fig. 1B). Heart rate and the high-frequency pulse interval component of the power spectra did not significantly alter throughout DH, suggesting that cardiac vagal tone is unchanged. Similarly, the very low-frequency (VLF) component of the systolic blood pressure (SBP) power spectra, related to endocrine components, was not significantly different at any point of the experimental period. However, the low-frequency (LF) component of SBP showed an increase 2 days after the onset of DH, suggesting an increase in SNA (P < 0.05 compared to baseline; Fig. 1C). Finally, the high-frequency component indicative of the respiratory modulation of systolic blood pressure was not different.

Figure 1.

A, radiotelemetry reveals an increase in mean arterial blood pressure over the time course of 3 days of dehydration. The arrow indicates the point of fluid withdrawal. Open and filled bars at the top represent 12 h of light and dark, respectively. B, quantification of mean arterial pressure in animals 1 day before (EH) and 1 to 3 days after water deprivation (DH1, DH2 and DH3, respectively 1, 2 and 3 days of dehydration). C, LF of SBP is significantly increased after 2 days of fluid deprivation. *Different from baseline (one-way ANOVA followed by Dunette's, P < 0.05, n = 8).

We next directly compared tSNA in the in situ working heart–brainstem–hypothalamus preparations (Antunes et al. 2006) of both EH (290 mosmol perfusate) and DH animals (340 mosmol perfusate) as based on plasma osmolality measurements made after 3 days of DH. tSNA was increased in 3–day-DH rats compared to EH animals (EH: 11.6 ± 3.6 μV, n = 14; DH: 23.5 ± 6.3 μV, n = 16; P < 0.05, Student's t test; Fig. 2A and B). Correspondingly, DH rats exhibited a higher perfusion pressure than EH preparations (72.2 ± 1.7 mmHg vs. EH: 66.6 ± 1.9 mmHg; Fig. 2B inset; n = 7; P < 0.05) with equivalent concentrations of perfusate VP and perfusion flow rates.

Figure 2.

Increases in integrated sympathetic nerve activity (SNA) were recorded directly from the thoracic sympathetic chain (A and B) in 3 day DH (Ab) rats compared to EH (Aa). Raw and integrated (∫) activities are depicted. This was associated with a rise in perfusion pressure indicative of increased vascular resistance (B, inset). Sympathetic activity has a central respiratory modulation which was significantly enhanced in DH rats during the inspiratory (TI) but not expiratory (TE) times (C). D, phrenic triggered averages indicate raised sympathetic activity during the phrenic nerve activity (PNA; inspiratory time). Note the change in scale of both PNA and SNA between EH and DH rats. See Supplemental Fig. S1 for the increased phrenic minute ventilation data in DH relative to EH rats. *Different from DH rats (Student's t-test, P < 0.05).

As previous studies have shown an altered strength in the coupling between respiratory drive and sympathetic activity in hypertensive rats (see Simms et al. 2009; Toney et al. 2010), we performed phrenic triggered averaging of tSNA. There was a significant increase in tSNA during inspiration in DH rats (TI: EH, 10.9 ± 1.5 μV, n = 7; DH: 26.1 ± 5.4 μV, n = 5; P < 0.05, Fig. 2C and D), but not expiration (TE: EH: 23.5 ± 2 μV, n = 7; DH: 42.8 ± 7.2 μV, n = 5; NS, Fig. 2C and D). Interestingly, the amplitude and inspiratory duration, but not frequency of phrenic discharge, were both elevated in DH rats (Supplemental Fig. S1). This computed to a ∼60% increase in neural inspiratory minute ventilation in DH (220.1 μV² min⁻¹) relative to EH rats (132.9 μV² min⁻¹; P < 0.05).

To be certain that the changes in tSNA were due to chronic dehydration, not simply the acute effects of increased perfusate osmolality, we performed an additional in situ experiment in which we switched the perfusate to one of normal osmolality (‘isotonic’ perfusate, i.e. 290 mosmol (kg water)⁻¹). When 3–day-dehydrated rats were exposed to isotonic perfusate there was no change in the level of tSNA whether the hypothalamus was present or dis-connected by precollicular transection (P > 0.05, n = 7; Supplemental Fig. S2). Moreover, because of different control of lumbar vs. renal sympathetic outflows during dehydration in anaesthetized rats (see Stocker et al. 2005), we also assessed whether this was the case between thoracic and lumbar SNA when switching from hyper- to isotonic perfusate in DH rats; we could not detect a difference (n = 5, P > 0.07; Supplemental Fig. S3).

The locus of sympathetic drive switches from the hypothalamus to the medulla oblongata following dehydration in situ

To determine the relative levels of angiotensin II type 1 AT₁ receptor ‘tone’ in the in situ preparation, we applied Losartan, an AT₁ receptor antagonist. In EH rats (n = 5), Losartan administered into the perfusate, and subsequent pre-collicular transection (to remove the forebrain) reduced tSNA by −1.6 ± 0.4 and −3.7 ± 0.9 μV (or 20 ± 3 and 44 ± 2%), respectively (Fig. 3A and B, P < 0.001). In contrast, in DH rats (n = 6) Losartan, subsequent pre-collicular and then ponto-medullary transections failed to reduce tSNA (Fig. 4A and B and inset). These data suggest that AT₁ receptors play a part in generating sympathetic tone in the EH but not DH rat. However, in the DH rat transection at the medullary–spinal cord junction reduced tSNA by −20.6 ± 9.8 μV (P < 0.05; or 70 ± 8%, Fig. 4B and inset; P < 0.01,). These data suggest that the main locus of sympathetic drive switches from the forebrain to the medulla oblongata as a consequence of DH.

Figure 3.

A, representative traces from a EH rat of raw and integrated (∫) thoracic SNA in EH rats before (control) and after administration of Losartan, an angiotensin II type 1 receptor antagonist. This was followed by pre-collicular transection (to remove the hypothalamus). B, quantification of changes in SNA (% of basal) after the above procedures. *Different from control; ^†different from Losartan (RM one-way ANOVA followed by Student–Newman–Keuls, P < 0.05, n = 5).

Figure 4.

A, representative traces of raw and integrated (∫) thoracic SNA in DH rats before (control) and after sequential systemic administration of Losartan, pre-collicular, post-collicular, medulla–spinal cord transections (see inset in B for histological analysis of a representative animal) and lidocaine applied to the sympathetic chain. B, quantification of changes as decribed above. *Different from control (RM one-way ANOVA followed by Student–Newman–Keuls, P < 0.05, n = 6).

In order to identify the brainstem nucleus involved in the generation of sympathetic activity in DH animals, we reversibly inactivated neuronal activity by injecting the GABA_A receptor agonist isoguvacine into the cNTS (see Supplemental Fig. S4 for injection sites). Reversible inactivation of the cNTS significantly reduced tSNA in DH (n = 5), but not EH (n = 6) rats (DH −10.5 ± 3.7 μV or −33 ± 7%vs. EH 0.1 ± 1 μV or 1 ± 7%; Fig. 5Aa, P < 0.01, Student's t test). In addition, the elevated phrenic amplitude observed in the DH rats (relative to EH) was reduced to levels seen in EH (Fig. 5Ad). To test the specificity of these effects, the injection site was also placed in the AP and in the medial NTS; the changes in baseline tSNA were reduced by 2 and 11% from baseline, respectively (n = 2 each; data not shown), much less than the reduction observed in the cNTS.

Figure 5.

Aa, effect of isoguvacine microinjection into the cNTS on ongoing SNA in EH (n = 6) and DH (n = 5) in situ rats. Ab, effect of isoguvacine injection into the cNTS on ongoing SNA in DH rats previously transfected with Ad-CMV-eGFP (eGFP, n = 5) or Ad-CMV-DNFosB-IRES-eGFP (dnFosB, n = 5). Ac, effect of pre-collicular transection on the ongoing SNA in control EH rats (n = 5) compared to DH rats previously transfected with dnFosB (n = 5). Ad and e indicates the changes in phrenic amplitude before and after inactivating cNTS in EH and DH rats (d) and after chronic blockade of FosB in NTS (e). *Different from EH (Fig. 5Aa) or from eGFP (Fig. 5Ab). (Student's t test, P < 0.05). B, confocal images of immunocytochemical staining for myc and eGFP from two animals confirming both the expression of transgenes (eGFP and dnFosB) and site within the cNTS. Abbreviations: CC, central canal; Gr, gracile nucleus; cNTS, commissural nucleus tractus soliatrii. See Supplemental Fig. S4 for NTS microinjection sites.

In vivo confirmation that the locus of arterial pressure control switches from the hypothalamus to the medulla oblongata following dehydration

Since previous in vivo evidence had stressed the importance of the PVN for maintaining arterial pressure in DH rats studied under anaesthesia (Stocker et al. 2005, 2008), we performed analogous experiments in conscious rats to substantiate our in situ findings. In EH rats, muscimol injections into the PVN bilaterally lowered mean arterial pressure by −8 ± 2 mmHg (Fig. 6A, n = 5; P < 0.05), similar to that reported previously (Stocker et al. 2005). However, in contrast to the results of Stocker et al. (2005), a similar small fall in arterial pressure was achieved in DH rats (−8 ± 4 mmHg; n = 5; P < 0.05) that was not different to that observed in EH rats (Fig. 6A; P = 0.44; Supplement Fig. S5). However, in anaesthetized rats, inactivation of the cNTS in DH rats produced a substantial depressor response (−34 ± 4 mmHg, n = 10; P < 0.001), which was significantly greater than that observed in EH rats (Fig. 6A, −13 ± 4 mmHg, n = 10; P < 0.001; Supplement Fig. S5). Indeed, the time to reach the maximal fall in blood pressure was fastest in the DH cNTS rats (3.6 ± 0.5 min; P < 0.05) compared to all other groups (cNTS EH, 9.9 ± 2.3 min; PVN EH, 15 ± 2.2 min; DH, 12 ± 2.9 min; two-way ANOVA). Injection sites were confirmed histologically (Fig. 6B).

Figure 6.

The effects of inactivating neurons in the PVN and cNTS (A) of euhydrated (EH) and 2–day-dehydrated (DH) rats. Muscimol (100 pmol (100 nl)^-1) was microinjected into the PVN bilaterally (100 nl each side) of conscious rats (EH, n = 5; DH, n = 5) and measurement made 20 min later. In the cNTS, muscimol (100 pmol (100 nl)^-1) was microinjected (25–30 nl) into the midline of anaesthetized rats (EH, n = 9; DH, n = 10) and data analysed 20 min later. Representative site of injections in PVN and cNTS, which were marked with 1% Evans blue (using the same volume as muscimol injections), are depicted in Ba and Bb, respectively. Tissue was counterstained with Giemsa. Note the significant role played by cNTS relative to PVN in maintaining arterial pressure in dehydrated rats. n.s. not significantly different. ***Different between EH and DH rats in cNTS group (two-way ANOVA followed by Student–Newman–Keuls, P < 0.001); ^†††Different between DH cNTS and DH PVN rats (two-way ANOVA followed by Student–Newman–Keuls, P < 0.001). CC, central canal.

Dehydration induced transcriptome changes in the NTS

In order to identify genes that might be responsible for regulating sympathetic activity in DH animals, we carried out microarray analysis on the NTS and the adjacent AP. Gene lists have been compiled that, with a high degree of statistical confidence, represent comprehensive descriptions of the RNA populations expressed in the NTS of EH (16,703 genes; see Transcriptomic Supplemental Files 6A and 6B, p. S1) and DH (16,898 genes, Transcriptomic Supplemental Files 6A and 6B, p. S2) rats, and in the AP of EH (15,133 genes, Transcriptomic Supplemental Files 6A and 6B, p. S3) and DH (16,088 genes, Transcriptomic Supplemental Files 6A and 6B, p. S4) rats. Combination of these catalogues with previously described (Hindmarch et al. 2006) datasets from the PVN (here re-analysed in GeneSpring GX11; EH, Transcriptomic Supplemental File 6B, p. S5 and DH, Transcriptomic Supplemental File 6B, p. S6) produced a list of 19,432 genes that formed the basis for statistical comparisons.

Because of the anatomical proximity of the AP and the NTS, it was important to establish that the transcriptional profile of each tissue was discrete. Mean Pearson correlation coefficients (Fig. 7A) reveal a high correlation between EH and DH arrays from the same tissue, but a low correlation between arrays from different tissues. While the correlation between the PVN and the AP was the lowest, the correlation between NTS and PVN was higher than between NTS and AP despite the anatomical proximity of these tissues. Principle components analysis based on experimental condition and all 19,432 genes in the combined gene list revealed that the anatomically discrete regions (AP, PVN and NTS) cluster differently with respect to their transcriptional profile (Fig. 7B). Because of these stark transcriptome profile differences between the NTS and AP, we are satisfied that the dissection of these anatomically proximate structures has been highly accurate.

Figure 7.

A, mean Pearson correlation coefficients reveal a high correlation between EH and DH arrays from the same tissue, but a low correlation between arrays from different tissues. B, PCA identifies directions along which the variation is maximal. The plot of PCA resolves different microarray samples in terms of how similar or different they are within in a three dimensional space that represents three different component scores, one for each axis. PCA based on experimental condition and all 19,432 genes in the combined gene list reveals that the anatomically discrete regions (AP, PVN and NTS) cluster differently with respect to their transcriptional profile. The component scores show that the highest degree of variability (54% and 41%) exists between tissues whereas the variability between arrays in the same tissue has a low score (6%). C, Venn analysis of genes differentially expressed as a consequence of DH reveals very little intersection, suggesting that the different brain regions respond to DH by altering the steady-state levels of very different sets of transcripts. Gene lists are presented in the appropriate Supplemental File 6B.

Statistical comparison (Welch ANOVA, P < 0.05, Benjimini & Hochberg, Tukey-HSD) revealed 429 genes in the NTS (Transcriptomic Supplemental Files 6A and 6B, p. S7), 491 genes in the PVN (Transcriptomic Supplemental Files 6A and 6B, p. S8) and 484 genes in the AP (Transcriptomic Supplemental Files 6A and 6B, p. S9) that have expression values that are significantly different following DH. One of the genes identified by microarray analysis as being up-regulated in the NTS by 3 days DH was AP1 transcription factor JunD.

Sympathetic drive in dehydrated rats is dependent upon AP1 transcription factor-mediated plasticity in the cNTS

Based on both our transcriptomic analysis of NTS (see above) and previous evidence that FosB, a member of the AP1 transcription factor regulated proteins, was up-regulated in the NTS on dehydration (Ji et al. 2007), we hypothesised that AP1 transcription factor activity might be responsible for the dehydration-induced plasticity in this nucleus. AP1 transcription factor activity in the cNTS was inhibited selectively using an adenoviral vector expressing a FosB dominant negative (Ad-CMV-DNFosB-IRES-eGFP) 5–7 days prior to 3 days dehydration (Fig. 5B). In these animals, inactivation of the cNTS using isoguvacine did not reduce tSNA (−1.4 ± 1.5 μV or −6.9 ± 8.7% compared to −13.8 ± 2.8 μV or −49 ± 8% in 3–day-DH rats with Ad-CMV-eGFP in cNTS (Fig. 5Ab; P < 0.05, Student's t test) nor did it alter phrenic amplitude (Fig. 5Ae). Interestingly, in DH animals in which AP1 activity is inhibited in the cNTS (n = 5), tSNA could be decreased by pre-collicular transection (−10.2 ± 2.9 μv or −45 ± 8%; n = 5) as in EH rats (−7.8 ± 2.1 μv or −35 ± 6%; n = 5; Fig. 5Ac). There was no change in both the reflex inhibition (baroreflex; P > 0.5) or excitation (peripheral chemoreflex; P > 0.75) of tSNA in animals in which the FosB dominant negative protein was expressed in the NTS (n = 6) vs. both naïve rats (n = 12) and those in which eGFP was expressed (n = 5).

Discussion

The data presented indicate that dehydration results in an increase in arterial pressure in freely moving conscious rats as measured using non-invasive radio-telemetry and is comparable with data from previous studies that used tail-cuff or cannula-based recording techniques (Burnier et al. 1983; Woods & Johnston, 1983; Gardiner & Bennet, 1985). Power spectral analysis of these data suggested that this was a consequence of an increase in SNA. In agreement with this suggestion, increased tSNA was directly measured in an in situ working heart–brainstem–hypothalamus preparation from DH rats. Consistent with previous studies (Allen, 2002; Colombari et al. 2002; Akine et al. 2003), we found that tSNA is dependent on the hypothalamus in EH rats. However, following dehydration, there is a switch in control to the cNTS. This plasticity seems to be by mediated by AP1 transcription factors, and microarray analysis has revealed a unique set of genes that might be AP1 targets that mediate the switch.

Changing mechanisms and neuroanatomical loci between EH and DH rats in sympathetic activity generation

Our data from EH rats are consistent with previous studies: similar to the findings of Stocker et al. (2005) we found a small reduction in tSNA and arterial pressure when dis-connecting the hypothalamus from the ponto-medullary brainstem in situ or following muscimol inactivation of the PVN in vivo in EH rats. In addition, although Freeman and Brooks (2007) observed a decrease in arterial pressure in 48–h-dehydrated rats after the blockade of AT₁ receptors in the PVN, this decrease was very small and in agreement with our suggestion that AT₁ receptors appear to have a minor role in the maintenance of arterial pressure in the dehydrated state.

A surprising finding was the switch in neuroanatomical locus for the control of tSNA from the hypothalamus to the brainstem as a consequence of dehydration. Reversible inactivation of neuronal activity in the cNTS by isoguvacine significantly reduced tSNA in DH rats, suggesting that it is this nucleus that is primarily responsible for driving thoracic sympathetic activity and contributes to the maintenance of cardiovascular homeostasis following chronic fluid deprivation. Indeed, this was substantiated by our in vivo findings demonstrating large falls in arterial pressure on inactivating cNTS in DH but much less in EH rats. Interestingly, although there was a fall in arterial pressure on inactivating the PVN in conscious DH rats this was relatively small (∼8 mmHg) and not different to the response seen in EH rats or the EH rats reported previously by Stocker et al. (2005). However, the present data do not support those of Stocker et al. (2005) since we were unable to replicate the importance of the PVN in maintaining tSNA and arterial pressure in the DH rat. Indeed, our novel findings go against the current dogma for a predominant role of the PVN in regulating SNA (and arterial pressure) during water deprivation (e.g. Stocker et al. 2006; Freeman & Brooks, 2007). Our data (both in situ and in vivo) support the novel idea that the cNTS plays a predominant role in regulating arterial pressure in dehydration. The reason for this difference is not clear but may, in part, reflect our use of unanaesthetized dehydrated rats for our PVN studies. Further, the degree of dehydration and its time course may also be contributing factors. The low circulating blood volume, high blood viscosity, haematocrit and plasma osmolality could all alter peripheral afferent drive to the NTS in the DH rat. All told, these factors may tip the balance away from the PVN and toward the cNTS in control of SNA in dehydration.

A novel role for AP1 transcription factor activity in switching the response to dehydration from the PVN to the cNTS

It has previously been shown that 48 h of dehydration increases the expression of the AP1 transcription factor FosB in the cNTS (Ji et al. 2007). Unfortunately, FosB probes are not represented on the microarray used in this study. However, we noted a significant up-regulation in the expression of JunD, a potential FosB partner in the formation of functional AP1 herterodimers. We, thus, hypothesised that AP1 transcription factor activity might be responsible for the DH-induced plasticity in this nucleus. Indeed, blockade of AP1 activity by the virally mediated expression of a dominant negative mutant FosB isoform in the cNTS of dehydrated rats prevented the reduction of tSNA induced by injection of isoguvacine into the cNTS, but pre-collicular transection (to remove the hypothalamus) resulted in a decrease in SNA, as in EH rats. Therefore, it seems that blockade of AP1 transcription factor activity in the cNTS prevents the switch in the control of tSNA from hypothalamus to cNTS, under which circumstances sympathetic drive remains hypothalamic in origin.

Respiratory coupling of sympathetic activity is exacerbated in dehydration

We found that DH rats studied in situ had an increased inspiratory modulation of the tSNA, but this finding was not supported by spectral analysis of systolic blood pressure from conscious animals. This may reflect a technical difference in assessment of respiratory-SNA modulation between conscious and in situ rats and future in vivo experiments with direct SNA recordings are needed to resolve this question. Based on the in situ data, we propose that the increase in tSNA is partly related to an elevated central inspiratory minute ventilation of these animals (Supplement Fig. S1). Interestingly, PVN activity is known to alter arterial pressure and ventilation during dehydration (Kc & Dick, 2010). Previous studies from our group have described altered coupling between the central respiratory oscillator and sympathetic networks under conditions of hypertension either expressed as a change in breathing pattern evoking an additional sympathetic burst (Zoccal et al. 2008) or stronger central coupling (Simms et al. 2009, 2010). The latter was based on the finding of elevated respiratory-related sympathetic activity without a change in phrenic activity in the spontaneously hypertensive rat (SHR). Herein, the mechanism for increased inspiratory-related tSNA appears to be distinct from the SHR and reflects a general increase in central inspiratory drive and not, necessarily, increased coupling of the two systems. Nevertheless, the relevance of the increased inspiratory-related tSNA may be that it provides a mechanism by which SNA is being elevated in conditions of dehydration.

The stimulus for the increased central respiratory drive eludes us presently. It is unlikely that dehydration is affecting blood gas and pH levels that activate chemoreceptors to elevate respiration as both circulating carbon dioxide levels and pH in the in situ preparation are maintained within normal limits. Alternatively, enhanced respiratory drive may be mediated via hyperosmotic stimulation of circumventricular organs and excitation of downstream hypothalamic and brainstem circuits regulating respiration as part of a pattern of response to dehydration. Or, alterations in the osmotic environment around the glomus cells of the peripheral chemoreceptors and/or neurones of the central chemoreceptors leads to elevated excitability. There is evidence that manipulation of the sodium–proton exchanger, which would occur with elevated extracellular sodium ion concentration, can influence phrenic nerve activity in vivo (Wiemann et al. 2008; Pasaro et al. 2009). This implies that any change in local proton concentration would affect pH-sensitive ionic channels (e.g. acid-sensing ion channels), which can alter peripheral chemoreceptor sensitivity (Tan et al. 2010).

Dehydration evoked transcriptomic changes in the brain regions controlling sympathetic activity

Dehydration evokes transcriptome changes in the PVN (Hindmarch et al. 2006) that are thought to regulate and mediate hypothalamo-neurohypophyseal system function-related plasticity (Hatton, 1997; Theodosis et al. 1998). Similarly, we have used microarray analysis to describe changes in gene expression that accompany dehydration in the NTS, and also the anatomically proximate AP. We note that the transcriptome changes in the NTS as a consequence of dehydration are very different from those observed in both the AP and the PVN (Fig. 7C). Venn analysis reveals an intersection of just three genes that are significantly and commonly regulated between the hypothalamus and the brainstem following dehydration (Fig. 7C). These were: (i) retinol saturase that promotes apidogenesis (Schupp et al. 2009) whereas retinol saturase knockout mice are obese (Moise et al. 2010); (ii) vgf was shown to regulate energy balance (Salton, 2003) and modulate sympathetic outflow to regulate fat storage and energy expenditure (Watson et al. 2009); (iii) Ankle2 (ankyrin repeat and LEM domain containing protein 2) for which there have been no functional studies performed. The LEM domain is a conserved protein sequence of 43 amino acids. It can be defined as: LEM = LAP2, EMERIN, MAN1. However, overall, each brain structure studied responded differently to the same stimulus of dehydration, which presumably reflects the unique homeostatic roles of these different brain regions. We also hypothesize that genes that significantly alter their expression as a consequence of dehydration in the NTS are responsible for regulating and mediating the consequent SNA plasticity and, as such, are excellent candidates for further study as potential AP1 targets.

Conclusions

We have shown that blood pressure and tSNA increase following chronic dehydration. There is considerable evidence that PVN neurons support SNA and blood pressure during water deprivation (Stocker et al. 2005, 2006, 2008; Freeman & Brooks, 2007). Our data suggest that, in addition, the brainstem, particularly the cNTS, adopts a more prominent role in the control of tSNA as a consequence of neural structure switching plasticity mediated by AP1 transcription factors that alters the expression of a unique set of genes.

Glossary

Abbreviations

AP

area postrema

cNTS

commissural nucleus tractus solitarii

DH

dehydration

EH

euhydration

eGFP

enhanced green fluorescent protein

IRES

internal ribosome entry site

PVN

paraventricular nucleus

SBP

systolic blood pressure

SHR

spontaneously hypertensive

SNA

sympathetic nerve activity

tSNA

thoracic SNA

VP

vasopressin

Authors' present addresses

D. S. A. Colombari: Department of Physiology and Pathology, School of Dentistry, São Paulo State University (UNESP), Rua Humaitá, 1680, 14801-903, Araraquara, SP, Brazil.

E. Colombari: Department of Physiology, UNIFESP-EPM, Rua Botucatu, 862, São Paulo-SP, Brazil.

Supplementary material

Supplemental Figure S1

Supplemental Figure S2

Supplemental Figure S3

Supplemental Figure S4

Supplemental Figure S5

Supplemental Figure S6

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors