Intravenous hypertonic NaCl acts via cerebral sodium-sensitive and angiotensinergic mechanisms to improve cardiac function in haemorrhaged conscious sheep

Abstract

Acute NaCl loading as resuscitation in haemorrhagic hypovolaemia is known to induce rapid cardiovascular recovery. Besides an osmotically induced increase in plasma volume the physiological mechanisms of action are unknown. We hypothesized that a CNS mechanism, elicited by increased periventricular [Na⁺] and mediated by angiotensin II type 1 receptors (AT₁), is obligatory for the full effect of hypertonic NaCl. To test this we investigated the cardiovascular responses to haemorrhage and subsequent hypertonic NaCl infusion (7.5% NaCl, 4 ml (kg BW)⁻¹) in six conscious sheep subjected to intracerebroventricular (i.c.v.) infusion of artificial cerebrospinal fluid (aCSF; control), mannitol solution (Man; 75 mmol l⁻¹[Na⁺], total osmolality 295 mosmol kg⁻¹) or losartan (Los; 1 mg ml⁻¹, AT₁ receptor antagonist) at three different occasions. Man normalized (144 ± 6 mmol l⁻¹, mean ±s.d.) the increase in i.c.v. [Na⁺] seen after aCSF (161 ± 2 mmol l⁻¹). Compared with control, both Man and Los significantly (P < 0.05) attenuated the improvement in mean arterial blood pressure (MAP), cardiac index and mesenteric blood flow (SMBF) in response to intravenous hypertonic NaCl: MAP, rapid response +45 mmHg versus+38 mmHg (Man) and +35 mmHg (Los); after 180 min, +32 mmHg versus+21 mmHg (Man) and +19 mmHg (Los); cardiac index after 180 min, +1.9 l min⁻¹ (m²)⁻¹versus+0.9 l min⁻¹ (m²)⁻¹ (Man) and +0.9 l min⁻¹ (m²)⁻¹ (Los); SMBF rapid response, +981 ml min⁻¹versus+719 ml min⁻¹ (Man) and +744 ml min⁻¹ (Los); after 180 min, +602 ml min⁻¹versus+372 ml min⁻¹ (Man) and +314 ml min⁻¹ (Los). The results suggest that increased periventricular [Na⁺] and cerebral AT₁ receptors contribute, together with plasma volume expansion, to improve systemic haemodynamics after treatment with hypertonic NaCl in haemorrhagic hypovolaemia.

Rapid infusion of a small volume hypertonic NaCl (∼7.5%, 4 ml (kg BW)⁻¹) has been shown to improve cardiovascular and immunological function in states of hypovolaemia, sepsis or traumatic injury (Kramer, 2003). Although extensively studied, the basic physiological mechanisms of action are not known in detail. The cardiovascular effects have been attributed mainly to plasma volume expansion caused by mobilization of endogenous fluids along an osmotic gradient (Kramer et al. 1989; Rocha-e-Silva & Poli de Figueiredo, 2005). Although an increased intravascular volume obviously benefits the circulation under such conditions it does not, by itself, explain the magnitude of improvement in cardiac output and blood pressure. Walsh & Kramer (1991) studied the degree of plasma volume expansion with hypertonic saline and different concentrations of dextran in anaesthetized dogs and noted a volume-independent component for the increase in cardiac output. A direct positive inotropic effect of hypertonic saline on the heart was suggested but studies addressing this question have resulted in contradictory results with observations of increases (Schroth et al. 2006), no changes (Welte et al. 1995; Ogino, 2002) and decreases (Constable et al. 1994; Waagstein et al. 1995) in contractility. In addition, hypertonic saline causes tachycardia when infused both during normo- (Hanwell et al. 1972) and hypovolaemia (Tollofsrud et al. 1998). Spinal nerve blockade at the thoracic level abolishes the heart rate response and reduces the increase in cardiac output when studied in anaesthetized normovolaemic dogs (Kien & Kramer, 1989), suggesting a neural component in the cardiac response to hypertonic saline.

Increasing brain Na⁺ concentration is well known to cause neurohumoral and haemodynamic alterations. Intracerebroventricular (i.c.v.) infusion of hypertonic NaCl increases arterial blood pressure (Andersson et al. 1972), central venous pressure (Hjelmqvist & Gunnarsson, 1995), vasopressin release (Swaminathan, 1980) and sympathetic nerve activity (Chen & Toney, 2001) in normovolaemic animals. In hypovolaemia the cardiovascular effects are even more apparent, with major improvements in arterial pressure and stroke volume as well as a pronounced tachycardia (Hjelmqvist et al. 1992). Interestingly, the majority of the responses to central infusion of hypertonic NaCl in normovolaemic animals are blocked by concomitant i.c.v. infusion of the angiotensin II type 1 receptor (AT₁) antagonist losartan (McKinley et al. 2003) indicating that they are mediated by a brain angiotensinergic mechanism. An interaction between hypertonic NaCl treatment and the cerebral angiotensinergic system has also been suggested since i.c.v. saralasin (a non-specific angiotensin II receptor antagonist) decreased survival after hypertonic NaCl resuscitation (Velasco et al. 1990). The concept of a potential cerebral pathway by which intravenous hypertonic NaCl may exert some of its effects have, however, been rather neglected in discussions of physiological mechanisms of hypertonic resuscitation.

Thus, our hypothesis was that the cerebrally induced haemodynamic effects would be of importance also when hypertonic NaCl was administered intravenously. If so, it would be able to explain some of the non-volume effects of hypertonic NaCl in treatment of haemorrhagic hypovolaemia. To test this hypothesis we kept cerebrospinal fluid [Na⁺] at near normal levels during and after intravenous hypertonic NaCl infusion in haemorrhaged conscious sheep. Due to the general dependency on central angiotensinergic pathways for increased brain [Na⁺] to exert its effect, we also investigated if a possible cerebral component of hypertonic NaCl resuscitation was mediated via AT₁ receptors. The results obtained are the first direct evidence for a cerebral Na⁺-elicited and AT₁ receptor-mediated mechanism for the cardiovascular responses to hypertonic NaCl in haemorrhagic hypovolaemia.

Methods

Animals

The experiments were conducted according to the European Union directive 86/609/EEG and the European Council convention ETS 123. The experiments were approved by the regional ethics committee in Stockholm.

Seven Texel cross-breed ewes weighing on average 55 kg (range 47.5–62.5 kg) were housed individually in pens where they had free access to water and a salt block. Twice a day they were fed hay and 75 g commercial pellets. All experiments were started between 08.30 and 09.30 h, approximately 1–2 h after the latest intake of food. The minimum interval between different surgical preparations/experiments was at least 10 days. During the experiments the animals were standing, unrestrained, in a cage in their habitual environment. All intravascular catheterizations were made under local anaesthesia (lidocaine hydrochloride, 5 mg ml⁻¹) and the animals were allowed at least 60 min of rest before the experiments started.

Surgical preparation

Five months before the start of the study, carotid artery loops were created. The carotid arteries were dissected free and placed in cervical skin that was closed around the arteries, making them accessible for easy cannulation. In six sheep ultrasonic flow probes (Transonic System Inc., New York, NY, USA) were placed around the left renal, the right femoral and the cranial mesenteric artery (corresponding to the superior mesenteric artery in humans). The cords were tunnelled subcutaneously to reach a paralumbar position where the connector was sutured to the skin. Finally, all sheep were equipped with two guide cannulae placed with their tips above the lateral ventricle on each side for intracerebroventricular infusions via an inner cannula. All surgical procedures were performed under isoflurane anaesthesia (2.1–2.3% end-tidal concentration) induced by intravenous (i.v.) sodium thiopental (10 mg kg⁻¹). After endotracheal intubation, facilitated by muscle relaxation using succinylcholine (1 g kg⁻¹i.v.), the sheep were mechanically ventilated. Post-operative treatment for 2 days with buprenorphine (0.002 mg kg⁻¹i.m.) and benzylpenicillin (20 000 IU kg⁻¹)–dihydrostreptomycin (0.0025 g kg⁻¹) were made routinely.

Intracerebroventricular and intravenous infusions

Artificial cerebrospinal fluid (aCSF) was prepared containing (mmol l⁻¹): 150 Na⁺, 2.9 K⁺, 1.1 Ca²⁺, 0.9 Mg²⁺, 155 Cl⁻, 24 HCO₃⁻, 0.5 HPO₄²⁻–H₂PO₄⁻ and a pH of 7.4. Angiotensin II (Bachem AG, Switzerland) and the AT₁ receptor antagonist losartan (Merck Research Laboratories, USA) were dissolved in aCSF to 7 μg ml⁻¹ and 1 mg ml⁻¹, respectively. To lower cerebrospinal fluid (CSF) [Na⁺] without changing osmolality, an aCSF solution was adjusted to contain 150 mmol l⁻¹ mannitol and 75 mmol l⁻¹ NaCl (Man; measured osmolality 295 mosmol kg⁻¹). i.c.v. infusions were made by lowering a probe of suitable length into one of the permanently placed guide tubes. Access to the lateral ventricles was tested by gentle siphoning of an aCSF-filled tube. When free communication with the CSF was confirmed, the probe was connected, via a silicone tube, to a 5 ml syringe placed in an infusion pump (802 Syringe pump, Univentor, Malta). All i.c.v. infusions were made at a rate of 1 ml h⁻¹. For intravenous infusion, a 7.5% NaCl solution was created by adding NaCl to sterile water. The hypertonic saline was manually injected via a jugular vein at 4 ml kg⁻¹ over 15 min.

Haemodynamic measurements

Arterial blood pressure was measured via a cannula in one of the carotid arteries. To measure cardiac output (CO), mixed venous oxygen saturation (S_vO2), central venous pressure (CVP) and pulmonary arterial pressure (PAP), a balloon-tipped pulmonary artery thermodilution catheter (7.5F Swan-Ganz, Baxter Healthcare Corporation, USA) was inserted into the right jugular vein. By pressure guidance it was advanced through the right atrium and ventricle and placed in the pulmonary artery. Signals from the thermodilution catheter were fed into a Vigilance Edwards Critical Care Monitor (Baxter Healthcare Corporation) where the CO was calculated after injection of 3 × 10 ml of ice-cooled saline into the right atrium. S_vO2 was measured continuously. Saline-filled tubes connected the arterial cannulae and the CVP and PAP lumen of the pulmonary artery catheter to pressure transducers (DPT-6003, PVB Medizin Technik, BMBH, Germany).

The flow probes were connected to two dual channel flow-meters (T 208, Transonic System Inc., New York, NY, USA) and renal (RBF), femoral (FBF) and superior mesenteric blood flow (SMBF) were recorded continuously. All signals were transferred to a data acquisition system (MP150, BIOPAC Systems, USA) and digitized with a sampling rate of 250 Hz. The data were stored on a computer and a selected number of parameters were shown on-line.

Responses to i.v. hypertonic saline in normovolaemic animals

Three sheep were fitted with an arterial and a venous catheter. After 30 min of baseline haemodynamic recordings, 7.5% NaCl (4 ml kg⁻¹) was infused over 15 min. Thereafter blood pressure and heart rate were observed for 90 min. Venous blood (5 ml) was drawn before the hypertonic NaCl infusion, directly after and once every 30 min for the following 90 min. At the same time points CSF (300 μl) was sampled from one lateral ventricle by means of gentle siphoning.

Verification of AT₁ receptor blockade by losartan

To ensure that losartan actually blocked cerebral AT₁ receptors two sheep received an i.c.v. infusion of angiotensin II with and without pre-treatment with losartan. After insertion of an arterial catheter and a baseline recording period, a 20 min i.c.v. infusion of angiotensin II (7 μg h⁻¹) was started. When MAP had normalized, losartan was infused i.c.v. (1 mg h⁻¹) for 50 min. Thirty minutes into the losartan infusion the angiotensin II infusion was repeated.

Responses to hypertonic NaCl resuscitation after intracerebroventricular mannitol or losartan

To investigate the hypothesis that normalizing the CSF [Na⁺] or blocking periventricular AT₁ receptors would impair the cardiovascular recovery in response to 7.5% hypertonic NaCl after haemorrhage, six sheep were assigned to a cross-over study where they, at separate occasions (minimum interval, 10 days) and in random order, received i.c.v. aCSF (control), Los or Man in combination with haemorrhage and hypertonic NaCl resuscitation. Besides an arterial and a thermodilution catheter, a two-lumen dialysis catheter was inserted in the left jugular vein for blood withdrawal. Regional blood flow (RBF, FBF and SMBF) were measured continuously. After the 60 min recovery period, baseline cardiovascular measurements were recorded for 30 min. During this period, venous (17 ml) and arterial (1 ml) blood samples were collected. Then a venous haemorrhage was started by withdrawing blood at a rate of 1 ml kg ⁻¹ min⁻¹. After 25 min venous and arterial blood samples were collected again, cardiac output measured and the i.c.v. infusion (1 ml h⁻¹) of either aCSF, Los or Man started. To keep the sheep hypotensive another 10 ml kg⁻¹ of blood was removed at a constant rate during the following 60 min. After another set of blood samples and cardiac output measurements, 7.5% NaCl (4 ml kg⁻¹) was infused over 15 min. Blood samples were taken and cardiac output measured directly and every 60 min after the end of the infusion. After an additional 3 h of cardiovascular monitoring after resuscitation, the i.c.v. infusion was discontinued and the shed blood re-transfused. All catheters were removed from the sheep which were allowed to return to their pens. In the control and the Man groups CSF (300 μl) was collected from the contra-lateral ventricle to the i.c.v. infusion side, at the same time points as blood samples were taken.

In one other sheep, not surgically prepared with flow probes but with i.c.v. cannulaes and carotid loops, the same protocol was followed but the infusions instead consisted of intravenous aCSF and Los (1 mg h⁻¹). This was done to rule out the possibility that Los acted in the periphery instead of in the brain.

Blood, plasma and cerebrospinal fluid analyses

Venous blood samples were immediately divided into two portions and put in pre-chilled tubes, one containing heparin and one EDTA. The blood was centrifuged at 60 g and aliquots from the EDTA–plasma were frozen (−20°C) for later determination of vasopressin (AVP) and angiotensin II concentrations. The heparinized samples were used for measurement of haematocrit, plasma osmolality (Auto & Stat Om 6010 osmometer; Kagaku Co., Japan), sodium and potassium concentration (IL 943 flame photometer; Instrumentation Laboratories, Italy) and protein concentration by refractometry (Atago Co., Japan). CSF samples were analysed for sodium and potassium concentrations. Blood gas analyses (Opti Critical Care analyser (AVL, Georgia, USA) were immediately performed on the arterial blood samples. Plasma concentrations of AVP and angiotensin II were quantified using commercial radioimmunoassay kits (Peninsula Laboratories Inc., member of the Bachem Group, Bachem Holding AG, Switzerland).

Calculations and statistical analyses

Mean arterial blood pressure (MAP) was calculated as the average value of a blood pressure cycle using data acquisition software (AcqKnowledge 3.8.1.; Biopac Systems). Heart rate (HR) was measured from the arterial blood pressure curve. Cardiac output was indexed to body surface area [0.09 × BW^0.67] (Graham et al. 1959) and presented as cardiac index. Total peripheral resistance (TPR) was calculated as [(MAP − CVP)/CO] and the other vascular resistances as [(MAP − CVP)/blood flow]. Stroke volume was computed as [CO/HR].

All statistical calculations were performed using Statistica 7.1 (Statsoft Inc.) and the graphs were created with Sigma Plot 8.02 (SPSS Inc.). The results are presented in the text as means and standard deviation (s.d.) or means and 95% confidence intervals (CI) with a significance level set as P ≤ 0.05. In the figures the results are presented as means and s.e.m. In the haemorrhage experiments data were analysed with two-way repeated measures analysis of variance (ANOVA; i.c.v. infusion × time) followed by Tukey's hsd test when suitable. In case of significant interaction between the variables, planned comparisons with Bonferroni-corrected results were made. Early effects of hypertonic NaCl were analysed between time points 85 and 100 min while later effects were analysed between time points 85 and 280 min. Differences in plasma AVP or angiotensin II levels between groups or over time were analysed with Friedman's ANOVA and Wilcoxon matched pair tests. The effects of hypertonic NaCl infusion in normovolaemic animals were evaluated with one-way repeated measures ANOVA followed by Tukey's hsd test.

Results

Effects of i.v. hypertonic NaCl in normovolaemic sheep

Hypertonic NaCl (7.5%, 4 ml kg⁻¹) rapidly increased MAP (+11 ± 4 mmHg) and HR (+14 ± 9 beats min⁻¹) (Fig. 1A). MAP remained elevated but HR decreased below baseline levels at the end of the 90 min observation period (−15 ± 4 beats min⁻¹). Assuming a constant total plasma protein amount during the experiment, plasma volume had increased by 22 ± 1% after the infusion (Fig. 1B). As the infused NaCl solution equilibrated with the extracellular fluid, plasma protein concentration returned towards pre-infusion levels and after 90 min the calculated plasma expansion was only 5 ± 3% compared with baseline. CSF [Na⁺] as well as plasma [Na⁺] increased from baseline levels (+5.2 ± 0.5 mmol l⁻¹ and +10.8 ± 2.5 mmol l⁻¹, respectively, Fig. 1C). The response in CSF [Na⁺] was slower compared with plasma [Na⁺], but remained well above baseline levels throughout the experiment.

Figure 1.

Effects of an intravenous (i.v.) infusion of 7.5% NaCl (4 ml kg⁻¹) in three normovolaemic sheep After 30 min of baseline registration, hypertonic NaCl was infused intravenously during 15 min. Blood and cerebrospinal fluid were sampled before hypertonic NaCl and every 30 min starting at the end of the infusion. Data are expressed as mean ±s.e.m. except in A where spreads are omitted to improve clarity. P, plasma.

The pressor effect of i.c.v. angiotensin II is blocked by losartan

Angiotensin II (7 μg h⁻¹) was infused i.c.v. in two animals and increased MAP by 15 and 12 mmHg, respectively. Losartan abolished this pressor effect when infused i.c.v. before and together with angiotensin II, indicating an effective central AT₁ receptor blockade. Figure 2 shows the original tracings from one of the sheep.

Figure 2.

Changes in arterial blood pressure in one sheep in response to i.c.v. infusion of angiotensin II with or without concomitant i.c.v. infusion of the AT₁ antagonist losartan Angiotensin II (Ang II) was first solely infused i.c.v. for 20 min (7 μg h⁻¹). When MAP had normalized the Ang II infusion was repeated 30 min into an i.c.v. infusion of losartan (1 mg h⁻¹).

Effects of haemorrhage and i.c.v. infusions of aCSF, mannitol or losartan before resuscitation

Considering that the sheep were subjected to repeated haemorrhage experiments, close attention was paid to behavioural, blood and cardiovascular parameters to assure that they had fully recovered from any previous experiment before starting another. There were some inter-individual differences in the acute responses to haemorrhage and the 60 min period of hypotension, but each sheep had a very consistent reaction, with only small differences in cardiovascular, hormonal or behavioural responses between the three occasions. No inter-group differences were observed in any variable until the onset of hypertonic NaCl infusion.

Haemorrhage resulted in a biphasic HR response. Initially, HR increased progressively in relation to blood loss to reach a maximum of 127 ± 22 beats min⁻¹, but when on average 14.1 ± 1.7 ml kg⁻¹ of blood had been removed HR fell abruptly. At the same time MAP, S_vO2, RBF and SMBF, which all remained more or less unchanged during the early phase of haemorrhage, were drastically reduced (data not shown), characteristic of the decompensatory phase (Schadt & Ludbrook, 1991). In some animals HR started to increase again after a period of relative bradycardia, while in some sheep it remained at near-baseline levels. However, analysed as a group, HR was significantly above baseline levels (P = 0.03, Fig. 3C) after removal of 25 ml kg⁻¹ of blood. MAP, cardiac index and CVP were reduced (P < 0.001, Fig. 3A, B and D) as well as all measured regional blood flow (P < 0.001, Fig. 4) and S_vO2 values (76.0 to 42.4%, P < 0.001). Haemorrhage significantly increased plasma [K⁺] (P = 0.005, Table 1), and reduced plasma protein concentration and the haematocrit (P < 0.001 and P = 0.02, respectively, Table 1). No significant changes were seen in blood gases but as the haemorrhage progressed the sheep started hyperventilating, leading to a tendency for reduction in P_CO2 (P = 0.10, Table 2). As expected, both plasma angiotensin II as well as AVP levels increased severalfold (P < 0.05, Fig. 5) in response to the initial 25 ml kg⁻¹ of blood loss. Peripheral vascular resistance estimated by calculated TPR was also significantly (P < 0.001, Fig. 6) elevated in all experimental groups.

Figure 3.

Systemic haemodynamic variables in six sheep subjected to haemorrhage and subsequent i.v. hypertonic NaCl in relation to different i.c.v. infusions Baseline registration was followed by haemorrhage at 1 ml kg ⁻¹ min⁻¹. After 25 min, the i.c.v. infusion (1 ml h⁻¹) was started and the haemorrhage rate decreased to remove an additional 10 ml kg⁻¹ for the following 60 min. At 85 min haemorrhage was stopped and 7.5% NaCl (4 ml kg⁻¹) infused i.v. for 15 min. The i.c.v. infusion was discontinued at 280 min. The i.c.v. infusion consisted of: control, artificial cerebrospinal fluid; losartan, AT₁ receptor antagonist, 1 mg ml⁻¹; mannitol, an iso-osmolar solution with 75 mm Na⁺ and 150 mmol l⁻¹ mannitol. Data are expressed as mean ±s.e.m. Significant (P < 0.05) differences in response to i.v. hypertonic NaCl compared with control are indicated by *(i.c.v. losartan) and # (i.c.v. mannitol). The analysis was performed between the time points indicated by the bar.

Figure 4.

Regional blood flow in six sheep subjected to haemorrhage and subsequent i.v. hypertonic NaCl in relation to different i.c.v. infusions Baseline registration was followed by haemorrhage at 1 ml kg ⁻¹ min⁻¹. After 25 min the i.c.v. infusion (1 ml h⁻¹) was started and the haemorrhage rate decreased to remove an additional 10 ml kg⁻¹ for the following 60 min. At 85 min haemorrhage was stopped and 7.5% NaCl (4 ml kg⁻¹) infused i.v. for 15 min. The i.c.v. infusion was discontinued at 280 min. The i.c.v. infusion consisted of: control, artificial cerebrospinal fluid; losartan, AT₁ receptor antagonist, 1 mg ml⁻¹; mannitol, an iso-osmolar solution with 75 mm Na⁺ and 150 mmol l⁻¹ mannitol. Data are expressed as mean ±s.e.m. Significant (P < 0.05) differences in response to i.v. hypertonic NaCl compared with control are indicated by *(i.c.v. losartan) and # (i.c.v. mannitol). The analysis was performed between the time points indicated by the bar.

Table 1.

Blood and cerebrospinal fluid (CSF) parameters in six sheep subjected to haemorrhage and subsequent intravenous hypertonic NaCl in relation to different intracerebroventricular (i.c.v.) infusions

		Baseline		25 min		85 min		100 min		160 min		220 min		280 min
	Group	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	P value
CSF Na+ (mmol·l−1)	Ctrl	150	1	—	150	1	161	2	159	1	158	1	157	1	—	—
	Los	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
	Man	149	2	—	—	136	6	144	6	142	7	144	5	145	4	< 0.001
Plasma Na+ (mmol·l−1)	Ctrl	148	1	149	2	147	2	164	2	156	2	155	2	154	1	—
	Los	148	2	147	1	148	5	166	2	159	4	158	3	157	3	n.s
	Man	146	1	147	1	147	1	165	4	156	2	156	1	154	2	n.s
Plasma K+ (mmol·l−1)	Ctrl	3.9	0.3	4.2	0.1	3.8	0.3	3.0	0.1	3.2	0.3	3.0	0.2	3.1	0.3	—
	Los	3.9	0.3	4.2	0.3	4.0	0.4	3.0	0.2	3.2	0.3	3.3	0.3	3.3	0.3	n.s.
	Man	3.9	0.2	4.3	0.2	4.0	0.3	3.0	0.2	3.2	0.2	3.0	0.1	3.0	0.2	n.s.
Plasma protein (mmol·l−1)	Ctrl	53	4	42	5	32	5	23	3	28	3	28	2	29	3	—
	Los	52	5	43	5	35	6	24	4	29	5	31	6	32	7	n.s.
	Man	52	2	43	3	34	4	26	3	29	3	31	1	31	1	n.s.
Haematocrit (%)	Ctrl	33	4	28	3	28	6	20	4	25	3	23	3	23	3	—
	Los	32	3	28	5	27	5	20	2	22	5	24	4	23	3	n.s.
	Man	33	3	30	4	30	4	20	2	24	3	23	4	24	3	n.s.

Blood samples were taken before haemorrhage (Baseline), after 25 ml kg⁻¹ of haemorrhage (25 min), after 35 ml kg⁻¹ of haemorrhage (85 min), directly after hypertonic NaCl (100 min) and once every hour for the following 3 h. The i.c.v. infusion of either a control artificial cerebrospinal fluid solution (ctrl), losartan (Los), or a low sodium, iso-osmolar mannitol solution (Man) started at 25 min and continued throughout the experiment. Data are expressed as mean and standard deviation (s.d.). The P value indicates the probability level for a difference compared with control in the response to hypertonic NaCl between baseline and 280 min. If the interaction effect between time and i.c.v. infusion was not significant (P > 0.05) this is stated as n.s. for both Man and Los.

Table 2.

Blood gases and acid-base status in six sheep subjected to haemorrhage and subsequent intravenous hypertonic NaCl in relation to different i.c.v. infusions

		Baseline		25 min		85 min		100 min		160 min		220 min		280 min
	Group	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	Mean	s.d.	P value
pH	Ctrl	7.49	0.03	7.52	0.06	7.53	0.02	7.40	0.02	7.48	0.03	7.53	0.02	7.55	0.03	—
	Los	7.50	0.03	7.52	0.04	7.55	0.07	7.42	0.04	7.47	0.02	7.49	0.01	7.49	0.04	< 0.001
	Man	7.49	0.03	7.51	0.06	7.54	0.07	7.41	0.04	7.52	0.04	7.56	0.03	7.57	0.04	0.75
PCO2 (kPa)	Ctrl	5.02	0.28	4.27	0.58	4.28	0.58	5.22	0.37	4.80	0.28	4.83	0.12	4.68	0.26	—
	Los	4.95	0.24	4.50	0.53	4.22	0.66	5.15	0.52	4.90	0.76	4.95	0.47	5.08	0.70	0.004
	Man	5.02	0.53	4.77	1.01	4.30	0.85	5.17	0.44	4.70	0.54	4.63	0.53	4.63	0.65	0.64
Base excess (mmol l−1)	Ctrl	4.80	2.39	3.22	2.46	4.00	3.56	−0.28	2.46	2.66	2.93	6.78	1.90	6.83	2.93	—
	Los	5.07	2.86	4.17	3.11	4.53	4.81	0.48	3.65	2.90	2.37	4.08	2.21	4.50	3.08	< 0.001
	Man	4.58	2.35	3.98	2.16	3.92	2.92	−0.38	3.12	5.25	2.22	7.82	2.22	8.13	2.07	0.55
PO2 (kPa)	Ctrl	16.0	1.3	16.3	0.5	16.7	1.5	15.6	1.2	15.7	1.3	15.7	0.9	15.8	0.9	—
	Los	16.5	0.8	15.3	1.0	16.8	1.5	15.3	1.1	15.2	1.3	16.2	1.5	15.9	1.4	n.s.
	Man	16.1	0.6	15.3	2.5	16.8	1.6	15.0	0.7	16.2	1.2	16.4	2.0	15.9	1.1	n.s.

Arterial blood was sampled before haemorrhage (Baseline), after 25 ml kg⁻¹ of haemorrhage (25 min), after 35 ml kg⁻¹ of haemorrhage (85 min), directly after hypertonic NaCl (100 min) and once every hour for the following 3 h. The i.c.v. infusion of either a control artificial cerebrospinal fluid solution (ctrl), losartan (Los), or a low sodium, iso-osmolar mannitol solution (Man) started at 25 min and continued throughout the experiment. 7.5% NaCl (4 ml kg⁻¹) was infused between 85 and 100 min. Data are expressed as mean and s.d. The P value indicates the probability level for a difference compared with control in the response to hypertonic NaCl between 85 and 280 min. If the interaction effect between time and i.c.v. infusion was not significant (P > 0.05) this is stated as n.s. for both Man and Los.

Figure 5.

Changes in plasma vasopressin and angiotensin II concentrations in six sheep subjected to haemorrhage and subsequent i.v. hypertonic NaCl in relation to different i.c.v. infusions Baseline registration was followed by haemorrhage at 1 ml kg ⁻¹ min⁻¹. After 25 min the i.c.v. infusion (1 ml h⁻¹) was started and the haemorrhage rate decreased to remove an additional 10 ml kg⁻¹ for the following 60 min. At 85 min haemorrhage was stopped and 7.5% NaCl (4 ml kg⁻¹) infused i.v. for 15 min. The i.c.v. infusion was discontinued at 280 min. The i.c.v. infusion consisted of: control, artificial cerebrospinal fluid; losartan, AT₁ receptor antagonist, 1 mg ml⁻¹; mannitol, an iso-osmolar solution with 75 mm Na⁺ and 150 mmol l⁻¹ mannitol. Data are expressed as mean ±s.e.m. For statistical calculations please refer to the Results section.

Figure 6.

Total peripheral resistance (TPR) in six sheep subjected to haemorrhage and subsequent i.v. hypertonic NaCl in relation to different i.c.v. infusions Baseline registration was followed by haemorrhage at 1 ml kg ⁻¹ min⁻¹. After 25 min, the i.c.v. infusion (1 ml h⁻¹) was started and the haemorrhage rate decreased to remove an additional 10 ml kg⁻¹ for the following 60 min. At 85 min haemorrhage was stopped and 7.5% NaCl (4 ml kg⁻¹) infused i.v. for 15 min. The i.c.v. infusion was discontinued at 280 min. The i.c.v. infusion consisted of: control, artificial cerebrospinal fluid; losartan, AT₁ receptor antagonist, 1 mg ml⁻¹; mannitol, an iso-osmolar solution with 75 mm Na⁺ and 150 mmol l⁻¹ mannitol. Data are expressed as mean ±s.e.m. No significant differences between the i.c.v. infusions in response to hypertonic NaCl were detected.

After the initial haemorrhage the i.c.v. infusion was started and blood removal continued at a slower rate for 60 min. The only difference between groups during this 60 min period was the reduction in CSF [Na⁺] over time in the Man animals compared with control (P < 0.001). Thus, Los or Man i.c.v. did not result in a change in the response to haemorrhagic hypovolaemia before the infusion of hypertonic NaCl commenced. The reduced haemorrhage rate allowed MAP, SMBF and RBF to recover slightly (P < 0.05) but cardiac index, CVP, S_vO2 and FBF did not change significantly (Figs 3 and 4, S_vO2 data not shown). Plasma protein concentration continued to fall (P < 0.001) but the reduction in the haematocrit did not progress further (Table 1). The hyperkalaemia was reversed and at the start of the hypertonic NaCl infusion there were no difference in [K⁺] compared with baseline (Table 1). Continuous hyperventilation during the entire haemorrhage period eventually induced a mild respiratory alkalosis (increased pH, P = 0.005; decreased P_CO2, P = 0.046, Table 2). Interestingly, although the haemorrhage continued, the elevated AVP levels declined (P < 0.05, Fig. 5). No significant corresponding change was seen in plasma angiotensin II concentration (Fig. 5).

Effects of maintaining CSF [Na⁺] or inhibiting brain AT₁ receptors on the responses to hypertonic NaCl resuscitation

Resuscitation with hypertonic NaCl (7.5%, 4 ml kg⁻¹) increased plasma [Na⁺] and CSF [Na⁺] (P < 0.001, Table 1). In the sheep receiving i.c.v. Man the reduced CSF [Na⁺] increased to near-baseline levels and remained there for the rest of the experiment. In all experiments the plasma [Na⁺] was still elevated at 280 min (P < 0.001 versus baseline). Thus, i.v. hypertonic NaCl induced a sustained rise in plasma [Na⁺] but in the Man group the CSF [Na⁺] was prevented from increasing above baseline levels.

Hypertonic NaCl infusion rapidly improved MAP and cardiac index in all groups (P < 0.001, Fig. 3), but there was also a significant interaction effect between time and i.c.v. infusion for both MAP (F_4,20= 5.88, P = 0.003) and cardiac index (F_4,20= 3.16, P = 0.036). Relative to control, planned comparisons revealed a reduction in MAP increase by 8 mmHg (17%) in the Man group (95% CI, 0.5–15 mmHg, P = 0.04) and 10 mmHg (23%) in the Los group (95% CI, 3–18 mmHg, P = 0.02) directly after hypertonic NaCl (Fig. 3). The attenuation of the effect remained throughout the observation period and at 180 min after hypertonic NaCl resuscitation, the animals receiving Man had 11 mmHg (33%) less increase in MAP (95% CI, 4–17 mmHg, P = 0.009) from pre-resuscitation values compared with control. The corresponding reduction in MAP effect for the Los group was 12 mmHg (39%, 95% CI, 6–19 mmHg, P = 0.005). No significant differences were observed for the immediate increase in cardiac index when comparing i.c.v. Man (P = 0.14) and i.c.v. Los (P = 0.24) to control (Fig. 3). However, the increase in cardiac index 180 min after hypertonic NaCl infusion was significantly reduced by 0.94 l min⁻¹ (m²)⁻¹ (50%) in the Man group (95% CI, 0.23–1.65 l min⁻¹ (m²)⁻¹, P = 0.02) and by 0.92 l min⁻¹ (m²)⁻¹ (49%, 95% CI, 0.34–1.50 l min⁻¹ (m²)⁻¹, P = 0.01) in the Los group. Hypertonic NaCl had no effect on TPR (P = 0.26) but this tended to be influenced by the i.c.v. infusions. However, there was no statistical significant interaction effect between time and i.c.v. infusion (P = 0.12, Fig. 6). In Fig. 7 the increase in cardiac index after hypertonic NaCl is illustrated in relation to the change in plasma volume (estimated by change in plasma protein concentration). On average, the control animals exhibited a greater improvement in cardiac index for each percentile increase in plasma volume.

Figure 7.

Changes in cardiac index after i.v. hypertonic NaCl (7.5%, 4 ml kg⁻¹) plotted against changes in calculated plasma volume in six haemorrhaged (35 ml kg⁻¹) sheep subjected to different i.c.v. infusions The cardiac index and plasma volume data are derived from 0 min, 120 min and 180 min after hypertonic NaCl and related to values directly before hypertonic NaCl was infused. Data from 60 min after hypertonic NaCl are omitted for clarity but included in the calculation of the regression lines (in total 24 observations in each line). The plasma volume changes are based on changes in plasma protein concentration. The i.c.v. infusion consisted of: control, artificial cerebrospinal fluid; losartan, AT₁ receptor antagonist, 1 mg ml⁻¹; mannitol, 75 mm Na⁺ in an iso-osmolar mannitol solution. Data are expressed as mean ±s.e.m.

Although the sheep already had tachycardia, HR increased further in response to hypertonic NaCl (P = 0.02, Fig. 3). However, the i.c.v. infusion × time interaction was not significant (P = 0.16). The calculated stroke volume also increased and remained elevated (P = 0.036, data not shown) with no obvious interaction effects (P = 0.32). CVP increased transiently (P = 0.02, Fig. 3) with no inter-group differences but soon fell back to pre-resuscitation levels. Hypertonic NaCl induced an increase in S_vO2 (P < 0.001) with a significant interaction between i.c.v. infusion and time (F_4,20= 3,7061, P = 0.020). No significant differences between groups could be seen immediately after hypertonic NaCl but at 280 min the Los group had recovered less (from 44.7 to 59.8%) than control (from 42.0 to 67.6%) (P = 0.02). The apparent reduction in improvement in the Man group (from 44.5 to 60.3%) compared with control did not reach statistical significance (P = 0.07).

The change in blood flow in the renal, mesenteric and femoral arteries was different in response to hypertonic NaCl infusion (Fig. 4). SMBF and FBF increased rapidly (P < 0.001 and P = 0.048, respectively) followed by a slow reduction. In contrast, RBF increased slowly (P = 0.001) during a longer period of time. There was no significant i.c.v. infusion × time interaction for RBF and FBF (P = 0.26 and P = 0.86, respectively) indicating no effect of the i.c.v. infusion on the change in blood flow in these arteries after hypertonic NaCl. In the mesenteric artery, however, the recovery in blood flow was significantly attenuated in sheep subjected to i.c.v. Man or Los both initially and 180 min after hypertonic NaCl (i.c.v. infusion × time interaction, F_4,20= 4.91 P = 0.009). The improvement in SMBF was reduced compared with control in the sheep receiving Man i.c.v. by 262 ml min⁻¹ (27%, 95% CI, 27–497 ml min⁻¹, P = 0.036) directly after hypertonic NaCl and 230 ml min⁻¹ (38%, 95% CI, 2–458 ml min⁻¹, P = 0.048) at 280 min. i.c.v. Los also reduced both the early and late improvement in SMBF to a similar degree: 236 ml min⁻¹ (24%, 95% CI, 42–430 ml min⁻¹, P = 0.027) and 288 ml min⁻¹ (48%, 95% CI, 243–332 ml min⁻¹, P < 0.001), respectively. There was no significant change in superior mesenteric resistance (SMVR) after hypertonic NaCl infusion.

Plasma [K⁺], protein concentration and the haematocrit decreased and remained below preresuscitation levels after hypertonic NaCl (P < 0.001 for all, Table 1). No effects of the different i.c.v. infusions on these variables were seen.

Initially, infusion of hypertonic NaCl significantly lowered pH (P = 0.001) and P_O2 (P = 0.01), and increased P_CO2 (P = 0.003) and base excess (P = 0.001) (Table 2). The changes were seemingly correlated to a decrease in respiration, but no exact measurements of respiratory rate or tidal volume were made. During the following 180 min pH (P < 0.001) and base excess (P < 0.001) increased again. Interestingly, there was a significant i.c.v. infusion × time interaction for both the change in pH (F_2,10= 8.27, P = 0.008) and base excess (F_2,10= 5.19, P = 0.028). Planned comparisons revealed a less pronounced increase in pH (P < 0.001) and base excess (P < 0.001) in the Los group compared with the other i.c.v. infusions (Table 2). This could, at least in part, be attributed to a different breathing pattern in the sheep receiving Los i.c.v., seen in the blood gases as an attenuated rise in P_CO2 (P = 0.004) compared with the other groups.

The haemorrhage-induced elevation of AVP and angiotensin II plasma levels were reduced after the hypertonic NaCl infusion (P < 0.05 for both, Fig. 5). In the Man group the reduction tended to be greater compared with the other i.c.v. infusions but variable individual effects resulted in a non-significant difference (P = 0.16).

In the experiment where losartan was infused intravenously, no difference could be seen compared with control in either cardiac index (rapid effect 4.7 versus 4.7 l min⁻¹ (m²)⁻¹, after 180 min 4.1 versus 4.3 min⁻¹ (m²)⁻¹) or MAP (rapid effect 95 versus 94 mmHg, after 180 min 80 versus 81 mmHg).

Discussion

The results of the present study support the hypothesis that hypertonic NaCl acts partly via cerebral mechanisms to improve cardiovascular function after haemorrhage. Counteracting cerebrospinal fluid increase in [Na⁺] as well as inhibiting central AT₁ receptors attenuated the improvement in cardiac index, MAP and mesenteric blood flow after hypertonic NaCl resuscitation. These observations suggest that (1) increased cerebral [Na⁺] at periventricular sites is necessary for the full haemodynamic effect of hypertonic NaCl and (2) central AT₁ receptors mediate hypertonic NaCl evoked recovery after haemorrhage.

Since no increase in TPR was seen after hypertonic NaCl in the control animals, the increase in MAP has to be caused by improved cardiac index. The reduced increase in MAP in the Man and Los groups could either be due to a more pronounced vasodilatory response or the attenuated improvement in cardiac index. The former seems unlikely since the tendency was rather an augmented increase in TPR in these groups compared with control (Fig. 6). In addition, the diminished blood flow in the mesenteric artery after i.c.v. Man or losartan is likely to be cardiac index related, since the reduction in SMBF corresponds well with the reduction in cardiac index. Alternatively, the lower MAP and cardiac index in the Man and losartan groups induced a reflex vasoconstriction of the splanchnic circulation. It is less plausible that the AT₁ receptor blockade or the normalization of CSF [Na⁺]per se caused vasoconstriction; if anything, the opposite would have been expected considering that hypertonic NaCl and angiotensin II i.c.v. increases TPR (Gunnarsson et al. 1994). Hence, our conclusion is that most of the differences in the haemodynamics seen between control and Man/losartan are due to the attenuated increase in cardiac index in these groups. Based on the fact that no decreased pre-load (estimated as CVP and plasma volume) or increased after-load (estimated as arterial blood pressure) by i.c.v. Man or losartan could be seen, the reason for the reduced cardiac index in the Man and losartan i.c.v.-treated animals ought to be a decreased inotropic and/or chronotropic effect. The results in the current study do not give enough information to establish the individual contribution of these factors since there were no significant differences in either HR or stroke volume compared with the control experiments. The interpretation is rendered even more difficult by the fact that a baroreceptor-mediated increase in HR and stroke volume would be expected to be more pronounced in the Man and losartan groups due to the lower MAP seen here. Thus, it is possible that both increased HR and contractility are the cause of the increased cardiac performance elicited by elevated central [Na⁺] and cerebral angiotensinergic mechanisms after hypertonic NaCl resuscitation.

Several observations indicate that the observed hypertonicity-induced increase in inotropy/chronotropy is neurally mediated. First, the tachycardia resulting from intravenous hypertonic NaCl can be inhibited by either β-adrenergic blockade or nerve blockade (Kien & Kramer, 1989). Second, i.c.v. hypertonic NaCl amplifies cardiac sympathetic nerve activity (CSNA) in conscious normotensive and normovolaemic sheep (Watson et al. 2004). This contributes to the [Na⁺]-induced hypertension and occurs in spite of baroreflex activation which counteracts the increase in CSNA. In hypovolaemic and hypotensive animals, as in this study, the increase in [Na⁺] would be expected to have an even larger impact on CSNA due to unloaded baroreceptors. A difference in HR response to hypertonic NaCl could also be seen between normo- and hypovolaemia with sustained tachycardia in haemorrhaged animals compared with a subsequent bradycardia in normovolaemic animals (Figs 1A and 3C). Intriguingly, studies reporting no change or decrease in contractility by hypertonic NaCl were mostly done during anaesthesia or in in vitro heart preparations (Hellyer & Meyer, 1994; Constable et al. 1994; Goertz et al. 1995; Waagstein et al. 1995; Welte et al. 1995; Ogino, 2002) while studies in conscious individuals usually report increased contractility after acute salt loading (Elgjo et al. 1998; Sirieix et al. 1999; Schroth et al. 2006). Anaesthesia has been shown to decrease CSNA (Matsukawa et al. 1993) and influence cardiac mechanical efficacy (Gare et al. 2001). In accordance with this we have recently shown an impaired ability to preserve cardiac output during haemorrhage by i.c.v. or intravenous hypertonic NaCl if administered to isoflurane anaesthetized sheep (Frithiof et al. 2006). Thus, it is possible that some anaesthetics attenuate the cerebral effects of hypertonic NaCl, explaining the inconsistent findings on cardiac performance after hypertonic NaCl.

i.c.v. infusions of hypertonic NaCl that are known to increase CSF [Na⁺] to levels seen in this study after intravenous hypertonic NaCl (i.e. 160–170 mmol l⁻¹) induce increases in arterial blood pressure (Andersson et al. 1972) and cardiac index (Gunnarsson et al. 1994) in normo- and hypovolaemic animals. The data in this study confirm that these central effects are obtained also when hypertonic NaCl is infused intravenously in haemorrhaged conscious sheep and do contribute to haemodynamic recovery. It is also indicated that this effect is achieved by increasing the CSF [Na⁺] rather than CSF osmolality. The intravascular osmotic gradient causing water removal from the brain was equally present also in the Man experiments and accordingly ought to have increased CSF osmolality similarly to control experiments. Intravenous administration of hypertonic NaCl in hypovolaemic sheep induced a sustained rise in CSF [Na⁺] that was more pronounced than when infused during normovolaemia (compare Fig. 1 and Table 1). The CSF [Na⁺] in the Man group was significantly lower than normal when the resuscitation with hypertonic NaCl started. That is, however, unlikely to have confounded our results since we saw no differences in the cardiovascular or hormonal responses in the Man experiments compared with the control experiments during the 60 min of i.c.v. infusion before resuscitation. Moreover, when previously tested, lowered CSF [Na⁺] and/or osmolality did not affect the cardiovascular response pattern or tolerance to haemorrhage (Ullman et al. 1993).

A cerebral increase in [Na⁺] also results in drinking and water seeking behaviour (Andersson et al. 1967), natriuresis (Mathai et al. 1998), vasopressin release (Swaminathan, 1980), inhibition of renal nerve sympathetic activity (RSNA) and renin release (May & McAllen, 1997), and elevated CSNA (Watson et al. 2004). These effects, together with the pressor response, are all abolished by i.c.v. losartan (Blair-West et al. 1994; May & McAllen, 1997; Mathai et al. 1998; Watson et al. 2004). Thus, cerebral angiotensinergic mechanisms facilitate many of the actions of hypertonic NaCl. The most probable explanation for this is stimulation of Na⁺-/osmo-sensitive cells (Voisin & Bourque, 2002) in the periventricular area leading to activation of angiotensin-containing neurons somewhere along the line to the autonomic and behavioural responses. It is possible that the fraction of improved haemodynamics achieved via increased cerebral [Na⁺] is also mediated via a central angiotensinergic pathway. The results obtained here with i.c.v. losartan are consistent with such a mechanism since the cardiovascular responses were the same as when CSF [Na⁺] was prevented from increasing above baseline values. However, there are some alternatives to consider. i.c.v. losartan does not alter baseline haemodynamics but has been shown to induce a premature hypotension when infused i.c.v. in conscious sheep subjected to blood loss (Mathai et al. 1997), indicating a central AT₁ receptor-mediated maintenance in arterial blood pressure during haemorrhage. Thus, we cannot completely dismiss the possibility that losartan interfered with neuro-cardiovascular responses, not stimulated by hypertonic NaCl, but by haemorrhage and hypovolaemia per se. Against this stands the fact that no attenuation in cardiovascular responses could be seen during the 60 min of i.c.v. losartan preceding resucitation. Furthermore, adding angiotensin II to a hypertonic NaCl i.c.v. infusion does not improve the tolerance to haemorrhage more than angiotensin II or hypertonic NaCl does by itself (Gunnarsson et al. 1994), suggesting a common pathway for hypertonic NaCl and angiotensin II in improving haemodynamics during haemorrhage. Another possibility is that losartan blocked the effect of circulating angiotensin II acting on peripheral receptors. We consider this unlikely based on (1) the observation that intravenous infusion of the same dose of losartan did not result in an impaired response to hypertonic NaCl and (2) reports of lack of peripheral AT₁ receptor inhibition after 180 min of 1 mg h⁻¹i.c.v. losartan (Dodic et al. 2006). In addition, during the 60 min of i.c.v. losartan prior to hypertonic NaCl, angiotensin II levels peaked but no effects were seen on MAP, cardiac index or regional blood flow, compared with control, indicating no major peripheral blockade of AT₁ receptors. That i.c.v. losartan in the administered dose does inhibit cerebral AT₁ receptors was confirmed by abolishment of the i.c.v. angiotensin II-induced MAP increase (Fig. 2).

Considering the Man and Los route of administration, Na⁺ and angiotensin II most probably exert their actions somewhere in the periventricular area. Lesion of the lamina terminalis, located in the anterior wall of the third ventricle, inhibits the increase in MAP and RSNA after i.c.v. hypertonic NaCl in conscious sheep (May et al. 2000). Interestingly, the same lesion also inhibits the MAP recovery after hypertonic NaCl resuscitation in anaesthetized rats subjected to haemorrhage (Barbosa et al. 1992). Of the three structures constituting the major part of the lamina terminalis (organum vasculosum laminae terminalis (OVLT), median preoptic nucleus (MnPO) and the subfornical organ (SFO)) the OVLT and SFO are known as sensory CVOs. These structures, lacking a functional blood–brain barrier have been shown to monitor changes in body fluid homeostasis and cardiovascular function (Cottrell & Ferguson, 2004). Neurons in the lamina terminalis are activated by hypertonic NaCl (Oldfield et al. 1991) and some use angiotensin II as a transmitter in their projections to the MnPO and the paraventricular nucleus of the hypothalamus (PVN) (Li & Ferguson, 1993). The lamina terminalis also contains a dense population of AT₁ receptors (McKinley et al. 2003) but considering that OVLT and SFO are CVOs they come in closer contact with circulating angiotensin II than the angiotensin II probably blocked by i.c.v. losartan here. It is thus plausible that a neural connection to the MnPO or the PVN mediates the neuro-cardiovascular effects of hypertonic NaCl if sensed at the lamina terminalis. That a polysynaptic connection to effector organs goes via the PVN is also reasonable since electrical stimulation of the SFO causes a pressor response that can be prevented by PVN lesion (Ferguson & Renaud, 1984). Moreover, recent studies have shown that PVN-projecting MnPO neurons respond to both hypertonic NaCl and angiotensin II with increased activity (Stocker & Toney, 2005) and that vasopressin-containing PVN neurons projecting to spinal preganglionic sympathetic neurons can mediate hypertonic NaCl-evoked sympathoexcitation (Antunes et al. 2006). The latter fits very well with the observation that a vasopressin V₁ receptor antagonist reduces the pressor effect of hypertonic NaCl in sepsis-induced hypotension in rats (Giusti-Paiva et al. 2007). Further evidence for a contribution of the sympathetic nervous system and the dependence of angiotensinergic pathways in this study arise from studies performed in normovolaemic, anaesthetized, sinoaortic baroreceptor-denervated rats (Chen & Toney, 2001). It was shown that a central increase in [Na⁺] induced an elevation in RSNA, slightly preceding the pressor response. Bilateral losartan injection in the PVN attenuated the increase in both RSNA and MAP. Taken together these data indicate that losartan in the current study may have blocked the transmission in the MnPO and/or the PVN. AT₁ antagonistic effects could also be achieved in the brainstem since nuclei there receive strong projections from PVN neurons (Dampney, 1994) and are also rich in AT₁ receptors (McKinley et al. 2003). Additional studies are warranted to investigate precisely the anatomical location(s) for the sodium–angiotensin II interaction in hypertonic rescuscitation.

The rapid improvement after hypertonic NaCl was similar in the mesenteric and the femoral circulation. However, unlike the SMBF, FBF quickly decreased to pre-resuscitation levels. This is probably a sign of a more prioritized blood flow to the intestines compared with the muscles in sheep. Considering the sensitivity of the intestines to hypoperfusion (Holland et al. 2005), the finding that the cerebral component of hypertonic NaCl resuscitation determines a large part of the increased blood flow to the splanchnic organs appears highly relevant. The RBF did not respond as quickly to the hypertonic NaCl infusion as FBF and SMBF but in view of the high circulating levels of the potent renal vasoconstrictor angiotensin II, a similar response pattern as in the other regional circulations could perhaps not be expected.

Since the plasma volume increased, dilution may account for most of the change in AVP and angiotensin II levels. In addition, the release, at least in the case of AVP, seems to decline slightly towards the end of the haemorrhage. Cerebral increases of [Na⁺] are known to cause an increase in plasma vasopressin and a decrease in plasma renin concentration via an angiotensinergic mechanism (May & McAllen, 1997; Mathai et al. 1998) but the lengthy hypovolaemia and hypotension may explain why the same effects were not seen here.

A somewhat unexpected finding was that i.c.v. losartan significantly attenuated a respiratory alkalosis seen in both the control and Man groups. Since no detailed respiratory monitoring was performed we cannot conclude on how this effect was achieved. Angiotensin II, presumably acting at cerebral sites, stimulates ventilation in normal (Olsson et al. 2004) and hypertensive rats (Jennings & Lockett, 2000) as well as during hypotension in dogs (Ohtake et al. 1993). It is evident that also the hyperventilatory response to hypovolaemia in conscious sheep may involve angiotensinergic mechanisms.

Conclusion

Although the expanded plasma volume is crucial for the haemodynamic improvement after hypertonic NaCl resuscitation in haemorrhaged conscious sheep, there is a contribution from mechanisms elicited at the cerebral level. The effect is achieved via elevated periventricular [Na⁺] and central angiotensinergic pathways to increase cardiac output, MAP and mesenteric blood flow. Brain hypertonicity-induced increase in CSNA may explain the augmented inotropy and chronotropy seen after hypertonic NaCl in conscious individuals. More studies are warranted to investigate the cerebral sites of action for intravenous hypertonic NaCl, the cellular basis for sodium–angiotensin II interaction in hypertonic resuscitation and also to establish the contribution of the central nervous system for the effects of hypertonic NaCl treatment in other disorders.