Subfornical organ neurons projecting to paraventricular nucleus: whole-cell properties

Brain Research 921 (2001) 78–85 www.elsevier.com / locate / bres

Research report

Subfornical organ neurons projecting to paraventricular nucleus: whole-cell properties James W. Anderson, Pauline M. Smith, Alastair V. Ferguson* Department of Physiology, Queen’ s University, Kingston, ON, Canada K7 L 3 N6 Accepted 28 August 2001

Abstract The subfornical organ (SFO) has been repeatedly identified as a CNS site that plays a critical role in sensing multiple physiological variables of the ‘milieu interieur’ and, through efferent projections to other CNS sites, initiating physiological responses to change. Many recent in vitro patch-clamp studies have examined the cellular mechanisms underlying the sensory abilities of these specialized CNS neurons. The primary limitation of these studies, however, has been the inability to identify homogeneous groups of SFO neurons for such investigation. We report here the development of techniques to permit patch clamp recording from dissociated SFO neurons identified according to their in vivo projection site. SFO neurons were labeled by injection of fluorescently labeled, retrogradely transported microspheres into the hypothalamic paraventricular nucleus (PVN) 3 days prior to cell dissociation. Patch-clamp recordings from these SFO-PVN neurons revealed both sodium currents, potassium currents, action potentials, input resistance and membrane potential which were all similar to SFO cells prepared from animals with no prior tracer injection. Labeled SFO→PVN cells were also found to be osmosensitive and responsive to angiotensin II, suggesting specific functional roles for this anatomically defined group of SFO neurons. Intriguingly, our post hoc analysis also demonstrated that all labeled neurons demonstrated a unique electrophysiological profile dominated by a large transient potassium conductance such that the transient / sustained potassium current ratio, or degree of inactivation was, on average, greater than 4.0. Utilization of these tracing techniques to permit the in vitro recording from cells with known in vivo connections will permit study of intrinsic mechanisms that underlie physiological responses of anatomically defined populations of neurons.  2001 Elsevier Science B.V. All rights reserved. Theme: Cellular and molecular biology Topic: Staining, tracing and imaging techniques Keywords: Circumventricular organ; Patch clamp; Osmolarity; Angiotensin; Potassium current

1. Introduction Circumventricular organ (CVO) neurons in general and the subfornical organ (SFO) in particular are uniquely situated within the central nervous system (CNS) since they lie outside the normal blood–brain barrier, lack extensive afferent neural input and have extensive efferent projections to central structures involved in the control of body fluid balance [15,22,9], and cardiovascular regulation [2–6,18]. Anterograde and retrograde tracing studies have identified the primary efferent projections of the SFO to both the hypothalamus and the anteroventral third ventricle *Corresponding author. Tel.: 11-613-533-2803; fax: 11-613-5336880. E-mail address: [email protected] (A.V. Ferguson).

(AV3V) region [14,15]. Areas of termination within the AV3V include the median preoptic area and the organum vasculosum of the lamina terminalis (OVLT) while hypothalamic projections include both the supraoptic (SON) and paraventricular (PVN) nuclei [15,17]. Other efferent projections include the regions of the zona incerta, raphe nuclei [21], infralimbic cortex and the rostral and ventral parts of the bed nucleus of the stria terminalis [13,19]. The sparse afferent inputs to the SFO are derived primarily from the midbrain raphe [13], median preoptic nucleus and nucleus reunions of the thalamus [16], outer layer of the lateral division of the parabrachial nucleus [8] and nucleus tractus solitarius (NTS) [23,20]. Although the development and use of the patch clamp technique to study intrinsic membrane properties of neurons has greatly enhanced our understanding of neuronal

0006-8993 / 01 / $ – see front matter  2001 Elsevier Science B.V. All rights reserved. PII: S0006-8993( 01 )03093-1

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

processes within the central nervous system, the technique still has certain limitations when utilized to study whole cell properties of intact neuronal circuits or in in vitro slice preparations. In general, because of the relatively large dendritic trees in most neuronal circuits, the accurate recording of whole-cell currents is made problematic by the inherent problems associated with poor space-clamp in these cells. Because of the large distances involved in clamping all the dendrites, voltages during voltage steps are not well controlled and are not necessarily constant throughout the entire step. In vitro recordings from dissociated neurons minimize many of these inherent difficulties with in vitro slice recordings, although the general lack of dendritic processes in these preparations does remove a number of potentially important physiological processes. Another inherent downfall of the dissociated cell preparation is the lack of information concerning the in vivo connection of the cell. In our laboratory, the development of a dissociated SFO cell preparation [7] allows us to accurately and unambiguously record the intrinsic membrane properties of these isolated neurons. Importantly, these dissociated neurons retain their responsiveness to a variety of physiological stimuli such as angiotensin II [7] and osmolality [1]. Here we report the results obtained utilizing a technique that permits the in vitro recording from neurons in which the in vivo connectivity of the cell has been established. Fluorescently labeled microspheres, which are taken up at the synapses of CNS neurons and then retrogradely transported back to the cell bodies, were first injected into PVN. Subsequent whole-cell patch clamp recording from the fluorescently labeled, acutely dissociated SFO neurons indicates that SFO neurons that project to PVN demonstrate homogeneous electrophysiological and functional profiles.

79

(five of five rats) while, following non-PVN injections, no label was observed in SFO neurons (none of three rats). Maximum visible spread of fluorescence was 200 mm from the tip of the injection cannula.

2.2. Cell isolation Dissociated SFO neurons were prepared for electrophysiological recordings using methods which have previously been established to produce neurons with functional properties similar to those observed in vivo [12]. Rats (which had previously received microinjections of FluoSpheres into PVN) were decapitated and the brain rapidly removed to iced (48C) Hank’s buffer (nominally Ca 21 and Mg 21 free). Under the operating microscope, the SFO was dissected free of surrounding tissue and then placed in 2 ml of the same solution warmed to 378C containing 1 mg / ml trypsin. Following incubation for 20–25 min, the SFOs were first transferred to an iced (48C) Hank’s solution containing Ca 21 (1.3 mM), Mg 21 (0.9 mM) and 0.1% bovine serum albumin (BSA) (Sigma type A-6003, essentially fatty acid-free) and triturated utilizing a 1-ml tuberculin syringe and a 20-gauge needle to promote the release of individual neurons into solution. This cell suspension was then centrifuged twice (9003g for 5 min at 48C), and the resultant pellet resuspended in the BSAcontaining solution, plated onto plastic petri dishes and placed within a 5% CO 2 environment at 378C until the cells attached to the dish (1 h). The petri dishes were then further filled with Neurobasal A media (Gibco) containing antibiotics (100 U / ml penicillin / streptomycin) and 0.5 mM L-glutamine. The cells were maintained in the CO 2 incubator at 378C until use (1–4 days following isolation).

2.3. Experimental solutions 2. Materials and methods

2.1. Microinjection Adult (100–150 g) male Sprague–Dawley rats were anesthetized with sodium pentobarbital (65 mg / kg) and then placed in a stereotaxic frame for subsequent microinjection of 0.5 ml of a 1% solution of 0.04 mm latex microspheres labeled with fluorescence (FluoSpheres). Coordinates for the injections were determined based on standard stereotaxic techniques and were optimized to permit bilateral injections into the PVN. The coordinates utilized for these bilateral PVN injections were bregma 10.9 mm, midline 60.4 mm and ventral 27.4 mm. Following this procedure, animals were permitted between 4 and 7 days not only to recover but also for the retrograde transport of the microspheres to take place. A pilot study demonstrated that fluorescent label was observed in SFO neurons (in histologically prepared sections of SFO) following injection of fluorescent microspheres into PVN

For both current-clamp and voltage-clamp recording, the pipette solution contained, in mM: potassium gluconate, 130; ethylene glycol-bis(b-aminoethyl ether)-N,N,N9,N9tetraacetic acid (EGTA), 10; MgCl 2 , 1; N-2-hydroxyethylpiperazine-N9-2-ethanesulfonic acid (HEPES), 10; Na 2 ATP, 4; GTP, 0.1; pH 7.2 with KOH. The extracellular bath solution contained, in mM: NaCl, 140; KCl, 5; MgCl 2 , 1; CaCl 2 , 2; HEPES, 10; glucose, 10 and mannitol to reach 300 mOsm. Where indicated, and tetrodotoxin (TTX) (1 mM) was added to the extracellular fluid to block voltage-gated sodium channels. The osmolality of all internal and external solutions was measured using the freezing point depression method (Osmette S, Advanced Instruments). All chemicals, unless otherwise stated, were obtained from Sigma Chemical Co. (St. Louis, MO).

2.4. Electrophysiological methods The recording chamber was the 35-mm plastic petri dish in which the cells had been cultured and solution changes

80

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

were made by a gravity-fed perfusion system. This flow was adjusted to approximately 4 ml / min and was maintained constant throughout the entire recording period. Electrodes of 1–4 MV were pulled from TW150 glass (World Precision Instruments, MA) on a horizontal micropipette puller (Model P-97, Sutter Instrument Co.), fire-polished and filled with the appropriate solution. Whole-cell access was obtained by applying brief suction to the electrode following formation of high resistance (typically 3–10 GV) seals and whole cell recordings acquired in both current and voltage clamp mode. Cells were defined as neurons by the presence of at least 60 mV action potentials in response to a depolarizing pulse during current-clamp recordings, and / or the presence of large (.2 nA) transient voltage-dependent Na 1 currents observed in voltage clamp. Signals were amplified, collected and processed using an Axopatch 200B (Axon Instruments) amplifier, and a 1401plus A–D interface and Signal and Spike2 software from CED (UK).

2.5. Statistical analysis All data are presented as means6S.E.M. Comparisons between two groups were performed using a standard t-test while a one-way analysis of variance followed by a Tukey test was used to determine significant differences between data containing three or more groups.

3. Results In dissociated cell preparations where labeling was observed (80%), approximately 30% of viable (by visual inspection) neurons contained fluorescent microspheres indicating former projections to the PVN. In cell preparations where no labeling within the neurons was noted, preparations were either devoid of fluorescent microspheres (injection into the brain but not into PVN), or large quantities of non-specific label combined with label present in any choroid plexus epithelium were observed (suggesting ventricular injections). All data presented were obtained from cell preparations in which neurons were specifically labeled. Of the total of 75 SFO neurons that were recorded from, 23 (31%) were fluorescently labeled (Fig. 1) and therefore identified as having projected to PVN in vivo. Sixteen of these fluorescent cells were initially examined utilizing the current clamp mode of recording to measure membrane potential changes and action potential frequency in response to either current injection or exposure of the isolated neuron to physiological stimuli. At rest, labeled neurons had a resting membrane potential of 254.061.8 mV and, in response to a hyperpolarizing injection of current, exhibited a change in membrane potential corresponding to an average input resistance of 2.460.24 GV, values which were not significantly different from those

obtained in 24 non-labeled cells (253.961.4 mV, and 2.6660.22 GV) recorded under identical conditions (Table 1). Voltage clamp analysis of all 23 labeled neurons showed large rapidly inactivating inward currents (Fig. 1A) which could be blocked by 1.0 mM tetrodotoxin (TTX) and were indistinguishable from the sodium current observed in non-labeled SFO neurons. Additionally, qualitative analysis suggested that all labeled cells demonstrate both a rapidly activating rapidly inactivating (putative IA ), and a non-inactivating (putative IK ), outward potassium current as illustrated in Fig. 1A and summarized in the mean I–V plots presented in Fig. 1B. In contrast, many non-labeled cells (16 of 52) mainly expressed only the relatively non-inactivating (putative IK ), outward potassium current with little if any contribution of the putative IA . These observations suggested the possibility that the transient inactivating K 1 current may in fact be a unique electrophysiological property of SFO-PVN neurons. Quantitative analysis of these currents in labeled cells was performed using the whole-cell voltage clamp technique to isolate and measure voltage-dependent potassium currents in the presence of 1.0 mM TTX. Recording of whole cell currents from a holding potential of 280 mV revealed a family of potassium currents with both a rapidly inactivating and non-inactivating component (ITOTAL ) (Fig. 2A, upper panel) while subsequent recording of currents from a more depolarized holding potential (240 mV), which leads to a steady-state inactivation of IA , produced slowly activating K 1 currents with little or no inactivation (IK ) (Fig. 2A, middle panel). The subtraction of these currents from ITOTAL permitted the component of the transient outward current (IA ) which was inactivated at a holding potential of 240 mV to be visualized as shown in the lower panel of Fig. 2A. Average I–Vs shown in Fig. 2B represent the means from seven fluorescent cells in which the IK and IA were isolated in this manner. In the absence of an intact sodium current, the IA was first activated at 240 mV and showed rapid inactivation during the 250-ms depolarizing pulse while IK was first activated at approximately 220 mV and showed little inactivation during the 250-ms pulses (Fig. 2A, lower panel). In labeled cells (n523), the average peak current measured at a test potential of 210 mV was 20.43360.035 nA which showed a decay to a mean current of 0.10560.011 nA (Table 1). Thus, the degree of inactivation (peak / steady-state current) present in this population of cells at 210 mV was 4.5560.37 (Table 1) as compared to a value of 3.1060.24 (n552, P,0.05) for all non-labeled cells. Interestingly, these non-labeled cells could be further separated out in that one group exhibited both a much smaller current at 230 mV and little inactivation at 210 mV as compared to either SFO-PVN or SFO-IA neurons. Separation of the non-fluorescent cells into two groups based on the degree of inactivation (DI) revealed the presence of two distinct subgroups of non-fluorescent cells, one of which was characterized by a DI greater than

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

81

Fig. 1. Fluorescently labeled SFO neurons exhibit normal whole cell currents. (A) Whole-cell current in a labeled neuron recorded in 250-ms, 10-mV steps from 2100 mV to 130 mV from a holding potential of 280 mV. Vertical current scale51 nA, horizontal time scale5100 ms. Inset is a schematic describing the method and a picture of both fluorescently labeled SFO neurons and non-labeled SFO neurons (white arrows). (B) Average I–V relationship (n523) for all labeled cells. IPEAK activates positive to 240 mV in these intact cells while the ISUSTAINED , which represents the component of outwardly rectifying current in these cells which does not inactivate at positive potentials, does not show significant activation until the depolarizing test potential exceeds 220 mV.

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

82

Table 1 Summary of electrophysiological data from SFO neurons split into two groups based on the presence / absence of fluorescent label following PVN injection of Fluospheres

Fluorescent (n523) (SFO-PVN) Non-fluorescent (n552) Total Degree of inactivation ,2 (SFO-IK ) Degree of inactivation .2 (SFO-IA )

Current clamp measurements

Voltage clamp measurements

n

VM (mV)

IR (GV)

n

IPEAK (230 mV) (nA)

IPEAK (210 mV) (nA)

ISUSTAINED (210 mV) (nA)

Degree of inactivation (DI) (IPEAK /ISUSTAINED )

16

254.061.8

2.4060.24

23

0.19460.021

0.43360.035

0.10560.011

4.5560.37

24

253.961.4

2.6660.22

52

0.13660.017

0.36860.032

0.12660.012

3.2860.25

9

254.562.0

3.3360.21

16

0.02160.007

0.18860.044

0.11660.025

1.6260.06

15

253.561.9

2.2660.29

36

0.18760.019

0.44960.034

0.13060.013

4.0360.28

2 (at 210 mV) (SFO-IA ) (n536) with a mean degree of inactivation of 4.0360.28 and a current measured at 230 mV of 0.18760.019 nA, values which were not signifi-

cantly different from labeled cells (Table 1). This subgroup of non-labeled cells exhibited peak currents and a DI not different from SFO-PVN cells at all test potentials as

Fig. 2. Fluorescently labeled SFO neurons contain at least two components of outward current. (A) Dissection of two components of whole cell current (same cell as Fig. 1A) utilizing voltage-dependent steady-state inactivation. (Upper panel) Total outward current measured in this cell from a holding potential of 280 mV (ITOTAL ). (Middle panel) Component of current remaining when holding potential was raised to 240 mV from 280 mV (IK ). (Lower panel) Component of current inactivated by increasing holding potential from 280 to 240 mV and visualized by the digital subtraction of IK from ITOTAL . All records collected in the presence of TTX (1.0 mM). Vertical current scale5500 pA, horizontal time scale5100 ms. (B) Average I–V relationship (n57) for labeled cells in which dissection of currents was accomplished. Vertical current scale5100 pA / pF.

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

illustrated in the upper panel of Fig. 3. In the 15 of these cells in which current clamp recordings were obtained, mean resting potential was 253.561.9 mV and input resistance averaged 2.2660.29 GV, values which again were not different from labeled cells (Table 1). Conversely, non-labeled cells that exhibited a much smaller outward current at 230 mV (0.02160.007 nA) and a significantly lower DI (1.6260.06, n516) as compared to either SFO-PVN cells or SFO-IA cells (Table 1) appear to represent a separate subpopulation of SFO neurons in which the IK is the dominant potassium conductance

83

(SFO-IK ). In current clamp, these cells exhibited a mean resting potential of 254.562.0 mV (P.0.05 as compared to either SFO-PVN or SFO-IA cells) and an input resistance of 3.3360.21 GV which was significantly greater (P,0.01) than the input resistance for either labeled or non-labeled cells with the higher degrees of inactivation (Table 1). The peak current measured in this subgroup of non-labeled cells was lower at all test potentials below 20 mV as compared to either SFO-PVN or SFO-IA cells while the DI was significantly lower in SFO-IK as compared to both other cell groups at all test potentials (Fig. 3, lower

Fig. 3. A sub-group of non-fluorescent SFO neurons (SFO-IK ) exhibits different current properties as compared to either SFO-PVN or SFO-IA neurons. (A) Original current traces, normalized to peak and overlaid for SFO-PVN, SFO-IA and SFO-IK neurons. SFO-PVN and SFO-IA currents are not different while SFO-IK neurons express more slowly activating currents with a much reduced degree of inactivation. (B) Left panel: Plot of peak current plotted against voltage for SFO-PVN (j), SFO-IA neurons (d) and SFO-IK neurons (s). Peak current for the SFO-IK neurons is significantly lower than all other cells at voltages less than 20 mV (P,0.05). Right panel: Plot of the degree of inactivation (IPEAK /ISUSTAINED ) current plotted against voltage for SFO-PVN (j), SFO-IA neurons (d) and SFO-IK neurons (s). Difference current for the SFO-IK neurons is significantly lower than all other cells at all voltages positive to 210 mV (P,0.05).

84

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

right panel). Representative currents for all three groups of cells measured at 210 mV, normalized so that all peak currents are equal and then overlaid to permit comparison, are plotted in Fig. 3, lower left panel. SFO-PVN and SFO-IA cells exhibit whole cell traces that clearly have a major IA component and are not different from each other (Fig. 3) while the SFO-IK cells exhibit both a slower activation and much-diminished DI at 210 mV as compared to either the SFO-PVN or SFO-IA traces (Fig. 3). The SFO has been suggested to be the CNS site at which peripheral changes in both angiotensin II (ANG) and osmolarity may act to control vasopressin secretion as a direct result of influences on the excitability of SFO neurons that project to the PVN and SON [7,1]. Therefore, we examined the responsiveness of labeled SFO neurons to bath administration of ANG or increases in bath osmolarity. The effects of bath administration of ANG on the membrane potential and spike frequency of four fluorescently labeled SFO neurons were examined. All of these cells responded with increases in action potential frequency (12.860.7 spikes / s) and depolarizations (16.460.9 mV) as illustrated in Fig. 4A. Similarly, hyperosmotic stimuli were observed to increase action potential frequency (11.660.4 spikes / s) and depolarize (15.261.1 mV) all four fluorescently labeled cells tested as shown in Fig. 4B.

4. Discussion The experiments described here provide clear evidence that injection and subsequent retrograde transport of fluorescent latex microspheres from PVN to SFO permits in vitro recording of both whole-cell currents and physiological responses in SFO neurons that projected to the PVN in vivo. Whole-cell patch clamp recordings from these labeled SFO neurons revealed normal currents (a fast inward sodium current and at least two outwardly rectifying potassium conductances) and spontaneous action potential generation in current clamp recording with normal input resistance and membrane potential as illustrated in Table 1 [7]. Importantly, we were also able to demonstrate that these labeled PVN projecting SFO neurons responded to physiological stimuli such as ANG [7], and changes in extracellular osmolality [1] in a manner effectively identical to that which we have previously reported in unidentified dissociated SFO neurons. These observations establish that loading of SFO neurons with fluorescently labeled latex microspheres (FluoSpheres) does not interfere with their ability to respond to physiological stimuli with normal cellular responses and is in strict contrast to recent reports suggesting that SFO neurons backfilled with DiI from the SON apparently do not respond to these stimuli [11]. This conclusion represents an essential step in establishing the feasibility of routine use of fluorescent

Fig. 4. Fluorescently labeled SFO neurons respond to normal physiological stimuli. (A) Continuous recording of action potential in an SFO neuron during a 90-s exposure to ANG (10 27 ). (B) Continuous recording of action potentials in cell exposed to 270 mosmol followed by 330 mosmol.

J.W. Anderson et al. / Brain Research 921 (2001) 78 – 85

microspheres for retrograde labeling of neurons prior to patch-clamp analysis in vitro. In addition, our experiments also lead to the conclusion that SFO neurons that project to PVN are intrinsically sensitive to both ANG and osmolality. This represents the first definitive demonstration of such specific functional properties of anatomically identified populations of SFO neurons. Finally, utilization of this technique has permitted us to conclude that SFO neurons which project to PVN have a common electrophysiological ‘fingerprint’ since they universally exhibit the presence of a large transient outward potassium conductance (IA ) which is activated at relatively hyperpolarized potentials and is completely inactivated at more depolarized holding potentials. In contrast, a subgroup of SFO neurons which were never labeled (SFO-IK ) expressed a different electrophysiological fingerprint characterized by the absence of the transient potassium conductance with a relative preponderance of an outwardly rectifying potassium conductance (IK ) which is activated at more depolarized potentials and exhibits little inactivation during short depolarizing pulses. In accordance with these observations we have quantitatively defined the electrophysiological properties of SFO neurons by calculating the ‘degree of inactivation (DI),’ which is obtained by dividing the peak outward potassium current measured at 210 mV by the sustained current at the end of a 250-ms depolarizing pulse, for each neuron recorded. Effectively we have observed that only those cells with a DI of .2 projected to PVN while we have never recorded from cells with a DI of ,2 which were identified as having projected to PVN. It should also be emphasized that in these experiments we also recorded from SFO neurons that were not labeled by successful PVN injections, yet exhibited the same IA . A certain percentage of the SFO-IA neurons almost certainly represent SFO-PVN neurons that failed to take up the fluorescent microspheres and therefore did not exhibit fluorescence. In addition, we cannot rule out the possibility that these cells represent another group of cells that also express large amounts of IA but do not project to PVN and therefore would not be expected to exhibit fluorescence following PVN injections. Clearly, to distinguish between these two possibilities would require the systematic utilization of this labeling technique in all regions of the central nervous system where SFO neurons are known to project including other hypothalamic nuclei such as SON. Interestingly, previous studies have not reported such functionally homogeneous electrophysiological properties of SFO to SON neurons [10].

5. Conclusions Subfornical organ neurons which project to PVN form a homogeneous subpopulation in that, in addition to fast sodium and the slowly inactivating potassium conductance (IK ), they express the rapidly activating, rapidly inactivat-

85

ing, outwardly rectifying potassium conductance (IA ) and retain their ability to respond to physiological stimuli (changes in osmotic pressure and ANG) expected to influence SFO neurons. In the future, further utilization of these tracing techniques to permit the in vitro recording from cells with known in vivo connections should permit a careful description of how intrinsic membrane conductances found on neurons with different efferent connectivity either differ or are similar.

Acknowledgements This work was supported by a grant to A.V.F. from the Canadian Institutes for Health Research.

References [1] J.W. Anderson, D.L. Washburn, A.V. Ferguson, Neuroscience 100 (2000) 539–547. [2] G.F. DiBona, J. Am. Soc. Nephrol. 10 (Suppl. 11) (1999) S90–S94. [3] A.V. Ferguson, P. Marcus, Am. J. Physiol. 255 (1988) R855–R860. [4] A.V. Ferguson, P. Smith, Regul. Pept. 27 (1990) 75–85. [5] A.V. Ferguson, T.A. Day, L.P. Renaud, Am. J. Physiol. 247 (1984) R1088–R1092. [6] A.V. Ferguson, J.S. Bains, V.L. Lowes, Circumventricular organs and cardiovascular homeostasis, in: G. Kunos, J. Ciriello (Eds.), Central ¨ Neural Mechanisms in Cardiovascular Regulation, Birkhauser, New York, 1994, pp. 80–101. [7] A.V. Ferguson, R.J. Bicknell, M.A. Carew et al., Neuroendocrinology 66 (1997) 409–415. [8] G.B. Gu, G. Ju, Brain Res. Bull. 38 (1995) 41–47. [9] A.K. Johnson, R.L. Thunhorst, Front. Neuroendocrinol. 18 (1997) 292–353. [10] R.F. Johnson, T.G. Beltz, M. Jurzak et al., Brain Res. 817 (1999) 226–231. [11] R.F. Johnson, T.G. Beltz, M. Jurzak et al., Am. J. Physiol. 280 (2001) R1592–R1599. [12] G.R. Li, J. Feng, Z. Wang et al., Am. J. Physiol. 270 (1996) C500–C507. [13] R.W. Lind, Brain Res. 384 (1986) 250–261. [14] R.W. Lind, A review of the neural connections of the subfornical organ, in: P.M. Gross (Ed.), Circumventricular Organs and Body Fluids, CRC Press, Boca Raton, FL, 1985, pp. 27–42. [15] R.W. Lind, G.W. Van Hoesen, A.K. Johnson, J. Comp. Neurol. 210 (1982) 265–277. [16] R.W. Lind, L.W. Swanson, D. Ganten, Brain Res. 321 (1984) 209–215. [17] R.R. Miselis, R.E. Shapiro, P.J. Hand, Science 205 (1979) 1022– 1025. [18] L.P. Renaud, A.V. Ferguson, T.A. Day et al., Brain Res. Bull. 15 (1985) 83–86. [19] L.W. Swanson, R.W. Lind, Brain Res. 379 (1986) 399–403. [20] J. Tanaka, Y. Hayashi, S. Shimamune et al., Brain Res. 777 (1997) 237–241. [21] J. Tanaka, A. Ushigome, K. Hori et al., Brain Res. Bull. 45 (1998) 315–318. [22] M.L. Weiss, G.I. Hatton, Brain Res. Bull. 24 (1990) 231–238. [23] A.M. Zardetto-Smith, T.S. Gray, Neurosci. Lett. 80 (1987) 163– 166.