INTRODUCTION

TOP ABSTRACT INTRODUCTION METHOD RESULTS DISCUSSION REFERENCES

CIRCULATING ANG II, such as that synthesized during a depletion of body sodium, has been implicated in the generation of salt appetite (4, 11, 27, 33). This suggests that circumventricular organs (CVOs) in the forebrain may be involved in the transduction of the ANG II signal from blood to brain (9, 17, 20). Lesions of either of two forebrain CVOs with cellular elements, the subfornical organ (SFO) or organum vasculosum laminae terminalis (OVLT), may reduce salt appetite induced by sodium depletion (12, 22, 26, 32), although a negative finding has been reported for SFO (21). It is unclear whether ANG II binding at the SFO stimulates salt appetite directly or whether it does so indirectly by supporting high water intake (21, 22). If ANG II receptors in the SFO or OVLT are critical for the direct stimulation of salt appetite by peripheral ANG II, it would seem logical that microinfusion of ANG II into either nucleus should elicit salt appetite. Data from our laboratory demonstrate that ANG II in and near the OVLT causes salt appetite but that ANG II in the SFO does not (10, but see also Refs. 1 and 3).

The present study addressed two independent questions, either of which might account for the discrepancies between SFO lesion and infusion studies: 1) is salt appetite blunted by elevated mean arterial blood pressure (MAP) during infusions of ANG II into SFO and 2) is the connectivity between the SFO and OVLT essential for salt appetite to be generated during a stimulation of the OVLT with ANG II?

Elevated MAP suppresses thirst induced by exogenous injection or infusion of ANG II (5, 7, 29), and this appears to be mediated by baroreceptor feedback (23). Similar mechanisms may exist for suppressing salt appetite (28). Because a bolus injection of ANG II in the SFO causes a brief rise of MAP (15), we hypothesized that a sustained infusion of ANG II in the SFO may cause a sustained elevation of MAP that inhibits salt appetite. Blocking this rise in MAP during an infusion of ANG II in the SFO might then unmask the expression of salt appetite, as is typically seen with an infusion of ANG II in the OVLT or the third ventricle (III V; Refs. 2 and 10). This speculation is supported by findings that injections of ANG II in the SFO did provoke substantial saline intake if serotonergic receptors in the lateral parabrachial nuclei were blocked (3) or if adrenergic receptor subtypes in the lateral hypothalamus were pharmacologically manipulated (1). The serotonin receptors were presumed to reflect cardiopulmonary or arterial baroreceptor activities that usually inhibit water and sodium intake during conditions of normal body sodium and blood pressure (3, 31). Blocking those receptors before an ANG II injection into SFO presumably released that inhibition and promoted high saline intake.

Experiment 1 was designed to test the effects of a reduction of MAP alone or of a reduction of both MAP and blood volume on water and saline intakes during infusions of ANG II into the SFO or OVLT. MAP alone was reduced by combined injections of minoxidil, a vasodilator, and captopril, an angiotensin-converting enzyme inhibitor. MAP and blood volume were simultaneously reduced by injections of furosemide, a diuretic and natriuretic, and captopril. We predicted that 1) sustained infusions of ANG II at a dose previously used to provoke saline intake (10) would elevate MAP, 2) the rise in MAP would be greater during an infusion of ANG II into SFO than into OVLT, a site that does support high saline intake during ANG II infusions, and 3) hypotensive pharmacological treatments that prevent a rise in MAP would greatly increase saline intakes during infusions of ANG II into the SFO.

In experiment 2 we tested a second, independent explanation for how SFO lesions might reduce salt appetite even if the SFO is not an important receptor site for ANG II in the elicitation of salt appetite. The full expression of salt appetite may require both an activation of the OVLT by ANG II and also a specific efferent signal originating from the SFO. This signal may result from the osmotic function of the SFO or from some other processing that is not directly related to the detection of circulating ANG II. This hypothesis correctly predicts that salt appetite should be interrupted by a lesion either of the OVLT (12), because of the loss of its ANG II receptor site, or of the SFO (26, 32), because of the loss of its efferent signal. The hypothesis makes no prediction about the effects of an infusion of ANG II at the SFO but it requires that the efferent connectivity of the SFO be intact in order for ANG II in the OVLT to generate salt appetite. This was tested by infusing ANG II into the OVLT of rats having either a knife cut of the rostroventral connectivity of SFO or a sham cut. A previous study in rats with SFO knife cuts examined water intake instead of saline intake and used infusions of ANG II into the optic recess of the III V instead of into the OVLT (14). The cuts in that case reduced, but did not abolish, water intake (14). We hypothesized that the knife cuts might similarly blunt the saline intake after infusions of ANG II into the OVLT.

	METHOD

TOP ABSTRACT INTRODUCTION METHOD RESULTS DISCUSSION REFERENCES

Animals. The subjects were 140 male Long-Evans rats weighing 300-500 g from the vivarium of the Department of Psychology at the University of Washington. They were maintained on Harlan Teklad rat chow with free access to both tap water and a solution of 0.3 M NaCl for at least 1 wk before the experiments. The lights were on in the rat holding rooms for 12 h per day, and temperature was controlled at 23°C.

Procedure. All rats were fitted with a chronic, 26-gauge stainless steel intracerebral guide cannula targeted either at the SFO or OVLT by aseptic stereotaxic surgery. The rats were anesthetized with Equi-Thesin (0.35 ml/100 g ip), the skull was leveled between bregma and lambda, and a 2.5-mm hole was drilled near bregma. The midsagittal sinus was retracted briefly while the cannula was lowered to coordinates given in the procedures for the two separate experiments. Gentamicin, 0.2 ml im, and topical Betadine were administered to inhibit postsurgical infection.

Experiment 1: Reduced MAP or blood volume. The coordinates for the guide cannulas in OVLT or SFO were as follows. For OVLT, the guide was lowered 7.2 mm ventral to the midsagittal sinus at a point 1.2 mm anterior to bregma. For SFO, the guide was moved to a point 1.5 mm posterior to bregma, the nose bar was raised 10°, and the guide was lowered 4.3 mm ventral to the sinus. Injectors in experiment 1 were inserted flush to the end of the cannula guide as in our previous study (10).

Experiment 1A: Reduced MAP. The arterial catheters were connected to the pressure transducers, and subcutaneous injections of 3 ml/kg were given 10 min later. The injections consisted of vehicle alone [7.5% (vol/vol) propylene glycol and isotonic saline] or else of a solution containing 0.25 or 0.75 mg/ml minoxidil (Sigma Chemical, St. Louis, MO) and also 25 mg/ml captopril (Bristol-Myers Squibb, Princeton, NJ). The minoxidil doses were either 0, 0.75, or 2.25 mg/kg, and the captopril dose was 75 mg/kg. Blood pressure was allowed to drop for 20 min after the minoxidil and captopril injections, at which time the intracranial injectors were inserted into the guide cannulas. An infusion of ANG II was started about 2 min after injectors had been inserted and about 30 min after the subcutaneous injections.

Experiment 1B: Reduced MAP and blood volume. This experiment was conducted for two reasons. First, minoxidil in experiment 1A might have had unintended suppressive effects on drinking and therefore we required a different procedure that lowered MAP without using minoxidil. Second, a reduction of blood volume in addition to MAP would reduce activity of both high-pressure baroreceptors and low-pressure volume receptors, either of which could inhibit saline intake during an infusion of ANG II into the SFO.

Experiment 2: Connectivity. Surgical procedures in experiment 2 were similar to those of experiment 1 with the following exceptions. In experiment 2 only OVLT cannulas were used. The OVLT guide cannula was lowered 6.3 mm ventral to the sinus so that the injectors extended 1.0 mm beyond the tip of the guide to reach OVLT. With this modification, it was the 33-gauge injector instead of the 26-gauge guide that approached closest to OVLT, and the smaller profile of the injector was less likely to rupture the ependymal lining of the ventricle. After the OVLT guide cannula was secured in place, a transection of the neural fibers of the SFO was made in some rats using a tungsten wire knife sheathed in the needle of a 1-µl syringe (13, 16). The transection was aimed at the rostral wall of the III V ventral to the SFO and dorsal to the anterior commissure using coordinates 1.8 mm caudal to bregma and 5.0 mm ventral to the sinus. The wire was extruded in an arc for 2 mm rostrally and then rotated 90° to the right and 90° to the left of the midline. Sham-cut animals had the sheath lowered, but the knife was not extended. After at least 1 wk of recovery from surgery, injectors were inserted 1.0 mm beyond the tip of the guide and the rats received a 3-h infusion of ANG II as in experiment 1. The rats were then anesthetized and perfused while the injector was still in place. This procedure allowed us to locate precisely the tip of the injector at the time of the infusion. Other details of the histology were as in experiment 1.

Statistics. Unless otherwise noted, all data are displayed as the means ± SE. Latency data are presented as medians. Each dependent variable in experiment 1A was analyzed using a two-factor completely randomized ANOVA with cannula site (SFO, OVLT, III V) and dose of minoxidil (vehicle, 0.75 mg/kg, and 2.25 mg/kg) as factors. For MAP a within subjects variable representing the hours of infusion also was used. Unplanned multiple comparisons were made using Tukey's honestly significant difference test. Planned comparisons were made using the least-significant difference test if the relevant F ratio was significant and using the Bonferroni correction if the F was not significant. The dependent variables in experiment 1B were analyzed using ANOVA for two independent samples. For MAP and heart rate, a within- subjects variable representing the hours of infusion was also used. Intakes in experiment 2 were analyzed by Student's t-test for independent samples.

	RESULTS

TOP ABSTRACT INTRODUCTION METHOD RESULTS DISCUSSION REFERENCES

Experiment 1A: Reduced MAP. After histology the sample sizes for each of the nine ANG II-infused groups included in the ANOVA were SFO-vehicle, 7; SFO-0.75 mg/kg, 9; SFO-2.25 mg/kg, 4; OVLT-vehicle, 6; OVLT-0.75 mg/kg, 9; OVLT-2.25 mg/kg, 5; III V-vehicle, 9; III V-0.75 mg/kg, 6; and III V-2.25 mg/kg, 5. Example photomicrographs of cannulas near the SFO or OVLT or in the III V are illustrated in Fig. 1.

View larger version (128K): [in this window] [in a new window]   Fig. 1.   Photomicrographs of sagittal sections of the tips of cannulas situated near the subfornical organ (SFO; A-C) or organum vasculosum laminae terminalis (OVLT; D-F) in experiment 1. Cannulas in A, B, D, and E are in tissue within 0.25 mm of the target nucleus. The median distance of all cannulas from the target was 0.1 mm. Cannulas farther away than those in D and E were excluded. Cannulas in C and F ruptured the ependymal lining of the third ventricle (III V). Bar = 0.5 mm.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Mean arterial blood pressure (MAP) during a 10-min preinfusion period (time 0) and throughout a 3-h infusion of ANG II into the SFO, OVLT, or the third ventricle (III V) in groups of rats pretreated subcutaneously with vehicle (Mx 0) or with 0.75 or 2.25 mg/kg minoxidil and 75 mg/kg captopril. MAP was sampled 100 times per second and averaged for each period.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Mean heart rate (HR) in beats/min during a 10-min preinfusion period (time 0) and throughout a 3-h infusion of ANG II into the SFO, OVLT, or III V in groups of rats pretreated subcutaneously with vehicle (Mx 0) or with 0.75 or 2.25 mg/kg minoxidil and 75 mg/kg captopril. Every beat was counted and averaged for each period.

View larger version (36K): [in this window] [in a new window]   Fig. 4.   Water intakes during a 3-h infusion of ANG II into the SFO, OVLT, or III V in groups of rats pretreated with vehicle (Mx 0) or 0.75 or 2.25 mg/kg minoxidil and 75 mg/kg captopril. Intake was significantly elevated by the lower dose of minoxidil only in the III V group.

View larger version (27K): [in this window] [in a new window]   Fig. 5.   Saline intakes during a 3-h infusion of ANG II into the SFO, OVLT, or III V in groups of rats pretreated with vehicle (Mx 0) or 0.75 or 2.25 mg/kg minoxidil and 75 mg/kg captopril. The SFO-infused groups failed to drink an appreciable amount of saline under any condition; the apparent increase at 0.75 mg/kg minoxidil resulted from 1 of 9 rats drinking 8.9 ml.

Experiment 1B: Reduced MAP and blood volume. The sample sizes of the captopril-treated groups after histology were 7 for the SFO-vehicle group, 11 for the SFO-furosemide group, 7 for the OVLT-vehicle group, and 5 for the OVLT-furosemide group.

View larger version (26K): [in this window] [in a new window]   Fig. 6.   Water and saline intakes during a 3-h infusion of ANG II into the SFO or OVLT in groups of rats pretreated with 10 mg/kg furosemide (Furo) and 100 mg/kg captopril or with vehicle (Veh) and captopril. A reduction of MAP and blood volume with furosemide and captopril failed to elevate water or saline intake, and rats with OVLT infusions drank more saline than those with SFO infusions.

Experiment 2: Connectivity. The sample sizes after histology for good OVLT cannulas and complete knife cuts were 10 in the knife-cut group and 12 in the sham-cut group. Knife cuts of the afferent and efferent neural connectivity of the SFO had no effect on either water (P = 0.62) or saline (P = 0.81) intakes during infusions of ANG II into the OVLT (Fig. 7).

View larger version (19K): [in this window] [in a new window]   Fig. 7.   Water and saline intakes during a 3-h infusion of ANG II into the OVLT in groups of rats having either a sham cut or a wire knife cut of the afferent and efferent fibers of the rostroventral pole of the SFO. Intake was not affected by the knife cuts.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHOD RESULTS DISCUSSION REFERENCES

The two parts of experiment 1 were designed to test the hypothesis that either preventing a rise in MAP alone or else simultaneously preventing a rise in MAP and reducing blood volume could unmask a salt appetite when ANG II was infused into the SFO of rats. Contrary to our predictions, sustained infusions of ANG II at a dose previously used to provoke saline intake did not elevate MAP throughout the experiment; MAP was not greater during an infusion of ANG II into SFO than into OVLT or III V; and hypotensive pharmacological treatments did not greatly increase saline intakes after infusions of ANG II into the SFO. The fact that MAP was always similar between groups receiving infusions into OVLT or SFO indicates that different pressor responses between the two sites do not account for the large differences in saline intake.

In experiment 1A, the effects of intracranial ANG II on MAP were transient in vehicle-treated rats. MAP was elevated very early after the beginning of the infusion but not during the next 3 h. Heart rate was also increased slightly in vehicle-treated rats during the first 2 min, although it was significantly reduced during the last 2 h of infusion. These early changes in MAP and heart rate were probably secondary to sympathetic excitation and in some rats to drinking generated by ANG II. However, the present dose of ANG II and method of infusion did not generate a sustained hypertension regardless of the site of infusion and this was true for both the total MAP, which included all of the activity of the animals, and the minimum 2-min MAP, which presumably reflected rest or sleep.

The absence of sustained hypertension does not indicate that no differences existed between rats infused at SFO or OVLT. The initial sympathetic excitation and hypertension caused by ANG II in vehicle-treated rats might have provoked counterregulatory responses, such as the secretion of vasodilators, and the levels of these might have differed by site of infusion. These different counterregulatory responses might then have been responsible for differential suppression of salt appetite. If so, then a reduction in MAP by minoxidil and captopril to a level below that of untreated rats should eliminate the necessity for a counterregulatory control of MAP and allow the expression of salt appetite during an infusion of ANG II into the SFO.

The MAP of the minoxidil-treated groups did not exceed that of untreated rats either during the first 2 min of infusion, when MAP was still falling as a consequence of minoxidil injection or during the subsequent 3 h. This was true despite the fact that similar numbers of rats drank water or saline during the initial 2 min in the vehicle- and 0.75 mg/kg minoxidil-treated groups. Heart rate continued to rise as MAP fell in minoxidil-treated rats, suggesting that the counterregulatory responses were to hypotension rather than to hypertension. Nevertheless, the minoxidil-treated rats receiving an infusion of ANG II into the SFO did not increase their intake of saline. The data suggest that neither an increase in MAP nor an increased counterregulatory response to hypertension suppresses saline intake in rats receiving ANG II infusions at the SFO. Instead, the results replicate our previous observation that the SFO does not support salt appetite during infusions of ANG II (10).

By contrast, a reduction of MAP with minoxidil and captopril did increase both water and saline intakes when the ANG II infusions included the III V. This supports the opinion that the consequences of hypertension during an infusion of ANG II are inhibitory to drinking (5, 7, 29) or that mildly reduced pressure facilitates drinking. The fact that minoxidil treatments failed to enhance salt appetite when ANG II infusions were made into the OVLT is curious and perhaps troublesome. This may indicate that MAP interacts with saline intake during infusions of ANG II only when the infusions are made into the forebrain ventricles. On the other hand it could also indicate that systemically administered minoxidil interacts with CVOs outside the blood-brain barrier in some way that reduces water and saline intakes. Minoxidil is a potassium channel opener in vascular smooth muscle and does not cross the blood-brain barrier. Minoxidil has been demonstrated to have no effect on isolated neurons from the hypothalamic paraventricular nucleus (19), but we cannot be certain it did not affect neurons and decrease fluid intake in the present study. This was most obvious at the large dose of minoxidil, which reduced water and saline intakes either directly or by causing low MAP (minimum 2-min MAP below 80 mmHg).

Experiment 1B was conducted both to reduce MAP without using minoxidil and also to reduce blood volume. The latter reflects the concern that salt appetite may be inhibited by information from either high-pressure baroreceptors or low-pressure volume receptors, or both. Thus reducing both blood pressure and blood volume may be necessary to unmask saline intake during infusions of ANG II into the SFO. We borrowed a procedure from R. L. Thunhorst (personal communication, December 1, 1998), who found elevated saline intakes during ANG II infusions into the cerebral ventricles that began 1 h after treatment with 10 mg/kg furosemide and 100 mg/kg captopril. The physiological and endocrine changes that occur with this treatment induce a natriuresis of about 1 mmol in 1 h and a drop in MAP of about 10-20 mmHg over the next 2 h (28, 30). Our results demonstrate a similar drop in MAP during the 3 h of ANG II infusion and also a loss of body weight at 40 min that was only slightly smaller than the urine volume attributed to diuresis at 60 min in Thunhorst et al. studies (28, 30). Thus our treatments were effective in reducing both volume and pressure in rats receiving infusions of ANG II into the SFO or OVLT.

Nevertheless, the furosemide and captopril treatments failed to enhance saline intake during the infusions at either CVO. Intakes of water and saline were already quite large in OVLT-infused rats, totaling nearly 30 ml of fluid in 3 h, so increasing that intake may have been difficult. SFO-infused rats, however, drank a similar amount of water to the OVLT-infused rats without drinking an appreciable amount of saline regardless of treatment.

As in a previous infusion study from this laboratory (10), the data rule out the obvious concern that all of the infusions may have acted simply by diffusing into the nearby ventricles regardless of the targeted site. If infusions into the SFO or OVLT acted by diffusing into the ventricles, then all infusions should have produced the same results. The fact that very different saline intakes were provoked by infusions into SFO, OVLT, and III V demonstrates again that the effects at the CVOs were local rather than global in the brain. As also observed in our previous study (10), sustained infusions at the present dose in a given nucleus produced 3-h intakes that were indistinguishable as long as the cannulas were situated within 0.25 mm of the target nucleus. Intakes in both studies declined rapidly with more distant cannula sites, and those have been rejected from the analysis.

It should be noted that more rigorous criteria are necessary for satisfactory drinking and pressor responses to occur after bolus injections, rather than sustained infusions, of ANG II at the SFO (15). Therefore, one potential interpretation of the present experiments is that the distance of some of the cannulas from the SFO produced responses in SFO that were too weak to generate a salt appetite, especially because the ventral hippocampal commissure could have acted as a barrier to diffusion. This is probably not the case for two reasons. First, these experiments employed the highest dose from a previous paper, 80 ng/h, even though a dose of 10 ng/h also produced significant salt intake during OVLT infusions (10). Even if only a fraction of the dose reached SFO, it was still a rather large amount of ANG II as evidenced by the high water intakes. Any dose near OVLT that generated this amount of water intake produced saline intake as well (10). Second, the conclusions are the same even if only those animals with exact placements of cannulas in the middle of SFO are counted. In 10 such rats the average intakes were 14.0 ml of water and 0.5 ml of saline in 3 h. The total saline intake of all 10 of these rats, 5.3 ml, was exceeded by 16 individual rats with infusions near OVLT in experiment 1. The 10 rats with exact placements of cannulas in the OVLT in experiment 1 averaged 16.1 ml water and 9.4 ml saline intake.

Therefore, our data support the view that infusions of ANG II in the tissue near OVLT are more potent than infusions near the SFO for generating salt appetite. Infusions near OVLT might stimulate the OVLT itself, the overlying median preoptic nucleus, or both (discussed in Ref. 10). For example, by penetrating and perfusing more brain tissue, the ventral cannulas near OVLT may be more effective in generating salt appetite because they act simultaneously at the OVLT and at terminal sites from the SFO in the ventral median preoptic nucleus. Completely isolated infusions into either SFO or OVLT alone might be equally ineffective. The closest we have yet been able to come to achieving an infusion at OVLT without input from the SFO is the knife cuts of experiment 2, which demonstrated that input from SFO is probably not necessary for an infusion of ANG II at OVLT to produce salt appetite. The exception to this probability is if the critical outputs from the SFO necessarily stimulate postsynaptic receptors in the region of the most ventral median preoptic nucleus, in which case the OVLT infusions may be stimulating both sets of receptors simultaneously.

The present findings are seemingly at odds with the data of Colombari et al. (3). These authors found that a blockade of serotonergic receptors in the lateral parabrachial nuclei, which were presumably carrying information from baroreceptors or volume receptors, produced a large salt appetite when ANG II was injected as a bolus dose into the SFO. In the absence of a blockade of these serotonergic receptors, an injection of ANG II into the SFO had no stimulatory effect on saline intake. One might have expected that an unloading of pressure and volume receptors in the present study would have accomplished a similar end to eliminating the neural feedback from those receptors as in Colombari et al., but the present data do not confirm that expectation.

A possible explanation of the apparent contradiction between the Columbari et al. study (3) and ours could be that the serotonergic synapses that they blocked in the lateral parabrachial nuclei may represent some source of inhibitory control over drinking other than baroreceptors or volume receptors. That is, if there is a serotonergic input from the hindbrain to the lateral parabrachial nuclei that tonically inhibits drinking and salt appetite independently from any activity of the baroreceptors or cardiopulmonary receptors, then the predicted results on behavior of a blockade of that mechanism would be identical to what they found (3, 18).

The results of experiment 2 demonstrate that the OVLT is independent of the SFO in its ability to support salt appetite during sustained infusions of ANG II, because high saline intake was observed even when the connectivity through the rostroventral stalk of the SFO was severed by a knife cut. Such knife cuts are essentially equivalent to SFO lesions in that they abolish water intake and pressor responses to circulating ANG II (14, 16), and this was verified with our method of producing knife cuts in the present study. The experiment was conducted to examine the possibility that the OVLT may be unable to generate salt appetite during exposure to ANG II without intact connectivity to and from the SFO. If true, then SFO lesions would be expected to reduce salt appetite even if ANG II acting at the SFO was not necessary for salt appetite. The data demonstrate that this is not the case, at least with respect to ANG II infused directly into the OVLT. The hypothesis may still be correct with respect to more subtle interactions of the SFO with neurons in the OVLT that are sensitive only to blood-borne ANG II.

Perspectives