Interleukin 1β modulates rat subfornical organ neurons as a result of activation of a non-selective cationic conductance

Abstract

The circumventricular organs (CVOs) are ideal locations at which circulating pyrogens may act to communicate with the CNS during an immune challenge. Their dense vasculature and fenestrated capillaries allow direct access of these pyrogens to CNS tissue without impediment of the blood-brain barrier (BBB). One such CVO, the subfornical organ (SFO), has been implicated as a site at which the circulating endogenous pyrogen interleukin 1β (IL-1β) acts to initiate the febrile response. This study was designed to determine the response of rat SFO neurons to IL-1β (1 nM to 100 fM) using whole-cell current-clamp and voltage-clamp techniques. We found that physiological (subseptic) concentrations of IL-1β (1 pM, 500 fM, 100 fM) induced a transient depolarization in SFO neurons accompanied by a significant increase in spike frequency. In contrast, pharmacological (septic) concentrations of IL-1β (1 nM) evoked a sustained hyperpolarization. While depolarizations in response to IL-1β were abolished by treatment of cells with the IL-1 receptor antagonist (IL-1ra), hyperpolarizations were still observed. Voltage-clamp analysis revealed that the majority (85 %) of SFO neurons responding to IL-1β with depolarization (29 of 34 cells) exhibited an electrophysiological profile characterized by a dominant delayed rectifier potassium current (DI_K), a conductance that we also found to be reduced to 84.4 ± 3.3 % of control by bath application of IL-1β. In addition, using slow voltage ramps we demonstrated that IL-1β activates a non-selective cationic current (I_NSC) with a reversal potential of −38.8 ± 1.8 mV. These studies identify the cellular mechanisms through which IL-1β can influence the excitability of SFO neurons and, as a consequence of such actions, initiate the febrile response to exogenous pyrogens.

Fever is a physiological response to the presence of exogenous substances which causes immune activation, resulting in an elevation in the ‘set point’ at which body temperature is regulated. This immune challenge in the periphery triggers complex changes in metabolism and in immune, endocrine and CNS function (Kluger, 1991; Rothwell, 1991). The febrile response can be experimentally induced by the administration of lipopolysaccharide (LPS), which in turn stimulates blood monocytes, phagocytic cells of liver and spleen, and other tissue macrophages to produce endogenous pyrogens such as interleukin 1 (IL-1; Tilders et al. 1994; Ma et al. 2000). This particular cytokine, which is now recognized as a key initiator of the febrile response, appears to play a prominent role in signalling peripheral immune activation to the CNS (Rothwell, 1991; Dinarello, 1994).

There are three known molecules in the IL-1 family: interleukin 1α (IL-1α), interleukin 1β (IL-1β) and IL-1 receptor antagonist (IL-1ra). IL-1α and IL-1β are 17.5 kDa molecules that act as agonists and bind to both type 1 (IL-1RI) and type 2 (IL-1RII) IL-1 receptors. IL-1ra, which lacks one of the binding sites of IL-1α and IL-1β to the two IL-1 receptors (Evans et al. 1995), is a pure endogenous antagonist of IL-1 action (Hannum et al. 1990; Eisenberg et al. 1990), which, despite binding to the receptor, does not evoke physiological responses (Dripps et al. 1991). Studies have shown that IL-1β is a much more potent fever inducer than IL-1α (Busbridge et al. 1989) and is crucial in febrile and neuro-immuno-endocrine responses (Horai et al. 1998).

However, the route whereby IL-1β transmits information from the periphery to the CNS is unclear. IL-1β does not cross the blood-brain barrier (BBB) in significant amounts (Coceani et al. 1988) because it is a large hydrophilic peptide. Circumventricular organs (CVOs), which lack the normal BBB, represent a potential target at which IL-1β can act to influence the CNS. These anatomically unique CNS structures contain fenestrated capillaries that allow the diffusion of lipophobic substances from the peripheral circulation to CNS neurons. The subfornical organ (SFO) is one such structure, which is located midline, on the roof of the 3rd ventricle and is partly covered by the diverging branches of the choroid plexus (Dellmann & Simpson, 1979). Previous studies showing that SFO lesion (Takahashi et al. 1997) as well as microinjection of IL-1ra into the SFO (Cartmell et al. 1999) both result in significant reductions in the fever induced by LPS have implicated this CVO as an important site in the development of fever. In addition, systemic administration of IL-6 has been reported to result in enhanced fos expression in SFO neurons (Vallieres et al. 1997), supporting the hypothesis that cytokines influence the excitability of SFO neurons. This study was therefore undertaken to examine the effect of IL-1β on isolated SFO neurons using whole-cell patch-clamp techniques.

METHODS

All experiments were approved by the Queen's University Animal Care Committee in accordance with the guidelines of the Canadian Council for Animal Care. SFO neurons were dissociated as previously described (Ferguson et al. 1997). Male Sprague-Dawley rats (125-150 g) were decapitated using a guillotine, and the brains quickly removed and immersed in ice-cold Hanks' buffer (Ca²⁺ and Mg²⁺ free, 0.03 M sucrose, pH adjusted to 7.4 with NaOH). A tissue block containing the hippocampal commissure and SFO was dissected and placed in Ca²⁺- and Mg²⁺-free Hanks' balanced salt solution. The SFO was separated from all surrounding tissue using microdissection techniques, placed in Ca²⁺- and Mg²⁺-free Hanks' balanced salt solution containing 1 mg ml⁻¹ trypsin (Sigma), and incubated in 5 % CO₂/95 % O₂, at 37 °C for 30 min. These tissue samples were periodically triturated (using a tuberculin syringe fitted with a 20 gauge needle) during incubation. Cells were centrifuged and resuspended in Hanks' solution containing Ca²⁺ (1.3 mM), Mg²⁺ (0.9 mM) and 0.1 % bovine serum albumin (BSA; Sigma type A-6003, essentially fatty acid free), centrifuged and resuspended a second time leaving a final cell suspension which was plated on plastic culture dishes (Corning), to which isolated cells adhered rapidly. Dishes were placed in the CO₂ incubator and 2 ml of Neurobasal-A medium (Gibco) containing 100 U ml⁻¹ penicillin-streptomycin and 0.5 mM L-glutamine was added to the culture dishes. Recordings were obtained from neurons 1-4 days following dissociation.

Electrophysiological techniques

Whole-cell patch-clamp recordings were obtained using fire-polished micropipettes pulled using a P-87 Flaming-Brown pipette puller (Sutter). Tip resistances were 2–4 MΩ when filled with a solution containing (mM): potassium gluconate (130), Hepes (10), EGTA (10), MgCl₂ (1), Na-ATP (4) and GTP (0.1); pH adjusted to 7.2 with KOH. The control bath solution consisted of artificial cerebrospinal fluid (ACSF) of the following composition (mM): NaCl (140), KCl (5), MgCl₂ (1), Hepes (10), glucose (10) and CaCl₂ (2); pH adjusted to 7.4 with NaOH. Tetrodotoxin (TTX; Alomone) was stored at a concentration of 100 μM in 100 μl aliquots of distilled water at −70 °C and prepared on the day of experimentation to a final concentration of 5 μM. IL-1β (Peprotech Inc., rat recombinant, 98 % purity) was stored at a concentration of 100 nM in 100 μl aliquots of distilled water at −70 °C. IL-1ra (Serotec) was stored at a concentration of 100 μM also in 100 μl aliquots in distilled water at −70 ° C.

Whole-cell recordings were obtained from SFO cells defined as neurons by the presence of voltage-gated Na⁺ currents under voltage clamp and at least 75 mV action potentials in response to a depolarizing pulse from a holding potential of −55 mV during current-clamp recordings. Signals were amplified using a List EPC-7 amplifier, filtered with an 8-pole Bessel filter at 1 kHz, digitized using the CED 1401 Plus interface (CED, Cambridge, UK) at 5 kHz, and stored on computer for off-line analysis. Data were collected using the EPC/Signal (Episode based capture) or Spike2 (continuous recording) packages (CED). Junction potentials were corrected by subtracting the offset potential obtained following entry of the electrode into the ACSF bath from the membrane potentials recorded.

Cell classification

Voltage-clamp recordings obtained from all cells prior to testing with IL-1β permitted classification of recorded neurons based on the expression of voltage-gated potassium channels (Anderson et al. 2001). One population of SFO neurons, the majority of which project to the hypothalamic paraventricular nucleus (PVN; Anderson et al. 2001), exhibited a dominant transient outward potassium current (DI_A). A second population of cells (the anatomical projection site of which has not been identified), in which the I_A was either not observed or was considerably smaller, was classified in accordance with their expression of a dominant sustained potassium current (DI_K). SFO cells could be readily separated into these two populations by measuring the peak current and the steady-state current as illustrated in Fig. 1, with cells displaying a peak to steady-state ratio greater than 2 classified as DI_A cells and those with a peak to steady-state ratio less than 2 classified as DI_K cells.

Figure 1. Voltage-clamp recordings indicate the presence of two populations of SFO neurons.

Cells were held at −70 mV and currents were elicited by 250 ms voltage steps from −70 to +30 mV in 10 mV increments. The trace on the lower left depicts a typical dominant I_A cell (DI_A) and the trace on the lower right a typical dominant I_K cell (DI_K). The pie chart displays the ratio of DI_A (60 %) to DI_K (40 %) cells recorded. The single traces show the −10 mV step, where the peak (*) to steady-state (○) ratio was determined. As indicated in the table, cells with a peak to steady-state ratio greater than 2 at this −10 mV step were classified as DI_A cells; those with a ratio less than 2 were classified as DI_K cells.

Statistical analysis

All data are presented as means ±s.e.m. Comparisons between group means were performed using Student's t test, and between cell population responsiveness were carried out with the Chi-square test, with P < 0.05 interpreted as indicative of significant differences in all instances. In all cases, changes in membrane potential during current-clamp recording of > 2 mV were used as an arbitrary cut off for defining an ‘effect’ of IL-1β.

RESULTS

A total of 113 isolated rat SFO neurons were recorded from in either current-clamp or voltage-clamp mode. These cells had resting membrane potentials between −55 and −48 mV, elicited action potentials with amplitudes of 75–115 mV, and had input resistances between 0.85 and 2.2 GΩ. In some cases where either full recovery was observed following an initial IL-1β test or an initial low dose test had no effect, single cells were tested more than once.

Response to IL-1β

Continuous current-clamp recordings were obtained from 69 SFO neurons to determine whether IL-1β affected membrane potential or firing frequency. After maintaining a control period of stable baseline recording for at least 120 s, IL-1β was administered by bath perfusion for 120 s at concentrations ranging between 100 fM and 1 nM, followed by a return to ACSF. Based on the established criteria, 45 of 69 cells (65 %) were classified as being responsive to IL-1β.

In accordance with reports of picomolar concentrations of IL-1β in the periphery during initiation of the immune response (Ma et al. 2000), we began our experiments with the application of 10 pM IL-1β. This concentration produced a transient depolarization in 7 of 10 cells tested (Fig. 2C), with this group of responsive cells showing a mean depolarization of 5.7 ± 0.9 mV, which occurred 6–28 s following IL-1β administration, an effect that was often followed by a sustained hyperpolarization (−3.9 ± 0.7 mV n = 6, data not shown). Similar transient depolarizations were also seen in response to 1 pM IL-1β in 7 of 9 cells tested (mean depolarization 5.4 ± 1.2 mV, latency 43–100 s, Fig. 2B), an effect that was again followed by sustained hyperpolarization in five cells (−4.2 ± 1.1 mV, data not shown). A further reduction in the dose of IL-1β to 500 fM, while producing similar depolarizations in 8 of 12 cells tested (6.4 ± 0.9 mV, latency 55–208 s, Fig. 2A), did not cause the longer latency hyperpolarizations observed at higher concentrations. Similarly 2 of 4 cells responded to 100 fM IL-1β with a depolarization of 3 ± 1 mV.

Figure 2. Current-clamp recordings during application of IL-1β.

Current-clamp traces showing responses to 120 s application of IL-1β. A-C, transient depolarizations with recovery at low dose 500 fM, 1 pM and 10 pM IL-1β, respectively. D, transient depolarization followed by a sustained hyperpolarization at 100 pM IL-1β. E, sustained hyperpolarization at high dose 1 nM IL-1β. Dotted lines represent resting membrane potentials.

Increasing the concentration of IL-1β administered to SFO neurons to high picomolar and nanomolar levels would be expected to result in bath concentrations similar to those seen in the periphery during extreme sepsis (Ebong et al. 1999). In our experiments, increasing the dose of IL-1β to 100 pM resulted in a sustained (no return to baseline) hyperpolarization in 13 of 21 cells (−8.8 ± 1.0 mV, latency 110–300 s, n = 1 3), which, in 10 of these 13 responding neurons, was preceded by a transient depolarization (mean 4.4 ± 0.6 mV, latency 6–16 s) as illustrated in Fig. 2D. Further increasing the dose of IL-1β tenfold to 1 nM resulted in a similar sustained hyperpolarization in 8 of 13 cells tested (mean −5.1 ± 1.0 mV, latency 55–100 s), in this case without any preceding depolarization (Fig. 2E). While these data do not indicate a clear traditional dose-response curve as shown in Fig. 3, they do quite clearly suggest two separate effects of IL-1β on SFO neurons, depolarizing at physiological concentrations, with the peak of this effect probably blunted by the hyperpolarizing effects observed at pathological concentrations similar to those anticipated during sepsis.

Figure 3. IL-1β influences membrane potential and input resistance.

A, peak changes in membrane potential for neurons responding to IL-1β with a transient depolarization at low doses (100–500 fM), a transient depolarization followed by a sustained hyperpolarization at intermediate doses (1–100 pM), or a sustained hyperpolarization in response to a high dose (1 nM − note that very few cells demonstrated such hyperpolarization in isolation). Each bar represents the mean ±s.e.m. for group data with n values indicated in boxes. B, IL-1β also decreased input resistance in SFO neurons as illustrated in the V-I curves shown here, which were obtained from a single SFO neuron prior to (▪) and during (□) the depolarization observed in response to 10 pM IL-1β administration. C, histogram summarizing the effects of IL-1β on 8 responsive (depolarization) SFO neurons in which input resistance decreased from 815 ± 57 to 690 ± 57 MΩ during IL-1β application with a recovery to 800 ± 59 MΩ (**P < 0.005, paired t test, n = 8).

Input resistance

IL-1β also influenced input resistance in IL-1β-responsive SFO neurons, as determined by the application of hyperpolarizing pulses during control and response periods. There was a significant reversible decrease in input resistance to 84.3 ± 3.0 % of the original control value during the depolarizing response (n = 8, P < 0.005, Fig. 3C), but no consistent changes during prolonged hyperpolarizing responses. In some cases full voltage- current (V-I) relationships were obtained by applying successive hyperpolarizing pulses of decreasing magnitude (−25 to −5 pA) before and after IL-1β application, and measuring peak changes in membrane potential (n = 3), as illustrated in Fig. 3B. The reversal potential estimated from such V-I relationships was −50 ± 1.5 mV, suggesting potential effects on a non-selective cationic conductance (I_NSC).

Receptor specificity

IL-1ra is an endogenous competitive antagonist of IL-1β as it binds to the IL-1 receptor and does not promote signalling (Dripps et al. 1991). We next examined whether the IL-1ra (no effect on membrane potential in isolation, n = 10) would block either or both of the responses to IL-1β. Administration of 10 or 100 nM IL-1ra at the same time as application of 100 pM IL-1β blocked the depolarization normally observed in response to IL-1β (Fig. 4A), as 0 of 7 cells responded following such pretreatment (Chi-square, P < 0.05 compared to control). However, hyperpolarization was still seen in two cells, with a peak membrane potential change of −7.3 ± 0.3 mV occurring 114 and 170 s following initiation of IL-1β administration, as illustrated in Fig. 4A. The maintenance of the hyperpolarization in the presence of IL-1ra (Chi-square, P < 0.1 compared to control) implies that activation of the IL-1β receptor is not responsible for this response. These data do, however, suggest that depolarizations are IL-1β receptor mediated, a conclusion further supported by the additional observation that depolarizing responses to 500 fM IL-1β were not observed following pretreatment with 100 pM IL-1ra (n = 4), as shown in Fig. 4B (Chi-square, P < 0.05 compared to control).

Figure 4. Current-clamp recordings during application of IL-1ra and IL-1β.

A, current-clamp recording from an SFO neuron illustrating the lack of a depolarizing response to 100 pM IL-1β (grey bars) administered at the same time as bath application of IL-1ra (open bars). Notice that although the depolarization was not observed, a longer latency hyperpolarization did still occur in response to this dose of IL-1β. Hyperpolarizing pulses of 25 pA were given at 10 s intervals throughout this recording. Bi, current-clamp recordings from a second SFO neuron that showed no response to 120 s application of 500 fM IL-1β during IL-1ra treatment. ii, in contrast a second application of the same dose of IL-1β following wash-out of IL-1ra results in a depolarization. Dotted lines represent resting membrane potential.

Spike frequency

The majority of cells that depolarized in response to IL-1β also showed an increase in action potential frequency. There was an overall increase in spike frequency in responsive cells that displayed spontaneous, non-bursting spike profiles during application of 10 pM to 100 fM IL-1β. The spike frequency during the 60 s control period before application of the drug was 0.80 ± 0.25 Hz and this increased to 1.20 ± 0.37 Hz during the initial 3 min after IL-1β application (paired t test, P < 0.05, n = 8). Peak spike frequency (within any single 20 s bin during this 3 min period) for each cell also increased from 0.88 ± 0.23 Hz during the control period to 1.98 ± 0.52 Hz during IL-1β application (paired t test, P < 0.005, n = 8).

Characteristics of responsive cells

We also compared the effects of IL-1β observed in DI_K cells (mean peak to steady-state ratio of 1.39 ± 0.06, n = 37), to those recorded from DI_A cells (mean ratio 3.57 ± 0.34, n = 19). Within the depolarizing dose range (100 pM to 100 fM, n = 56), 29 of 37 DI_K cells (78 %) responded to IL-1β with depolarization, while only 5 of 19 (26 %) DI_A cells responded (Fig. 5), suggesting that the former subpopulation of SFO neurons respond preferentially to IL-1β. In total, responsive cells showed a mean peak to steady-state ratio of 1.73 ± 0.17 (n = 34), which was significantly different from that observed in non-responsive neurons (2.75 ± 0.35, n = 22, t test, P < 0.01).

Figure 5. Peak to steady-state ratios of IL-1β-responsive and non-responsive neurons.

Peak to steady-state ratios of cells within the effective IL-1β depolarizing dose range (100 pM to 100 fM); 34 of 56 cells depolarized, 29 of which were classified as DI_K as they had a peak to steady-state ratio less than 2. The mean values ±s.e.m. for the responsive and non-responsive cell groups (▪) were 1.73 ± 0.17 and 2.75 ± 0.35, respectively (*P < 0.01). The inset depicts the percentage of depolarizing cells in the DI_K (78 %, 29 of 37 cells) and DI_A (26 %, 5 of 19 cells) cell populations.

Non-selective cation channels

Voltage-clamp experiments were undertaken to determine the identity of the current underlying the effects of IL-1β observed in DI_K cells. Our input resistance data, as well as previous studies suggesting that depolarizing effects of a variety of peptides are mediated by the opening of non-selective cation channels (NSCCs; Oliet & Bourque, 1993; Takano et al. 1996; Washburn et al. 1999a), led us to examine the effects of IL-1β on this current, as assessed by the application of slow (12 mV s⁻¹) depolarizing voltage ramps (−80 to +20 mV; see Fig. 6A). Bath application of 500 fM IL-1β produced an increase in conductance over the voltage range in 9 of 15 DI_K cells tested (60 %), a proportion similar to that demonstrating depolarizing effects (Chi-square, P < 0.1). The subtracted current (Fig. 6A inset) was linear through the range −80 to −15 mV, indicating a non-rectifying current. The slope was 0.60 ± 0.12 pA mV⁻¹ and the mean reversal potential was −38.8 ± 1.8 mV (Fig. 6B). Recovery was seen in all cells following return to ACSF and the response was repeatable following subsequent similar IL-1β applications (n = 3). This current was not seen in response to 500 fM IL-1β in the presence of 100 pM IL-1ra (mean slope 0.013 ± 0.008 pA mV⁻¹, n = 4).

Figure 6. Voltage-clamp recordings displaying activation of NSCC following IL-1β application.

A, voltage-clamp recordings showing the currents produced by a 12 mV s⁻¹ depolarizing ramp. Shown here is the control current, the current elicited by 500 fM IL-1β (red) and the recovery current following ACSF wash. The difference current obtained by subtracting control currents from currents measured during IL-1β administration are displayed in the inset. B, average linear difference current obtained from the 9 cells (out of 15) that responded to 500 fM IL-1β during this slow ramp protocol. The slope of this current was 0.60 ± 0.12 pA mV⁻¹ and the reversal potential was −38.8 ± 1.8 mV.

Potassium currents

The above slow ramps also revealed that the difference current between control and IL-1β application was non-linear in the upper depolarized voltage range (0 to +20 mV), suggesting the influence of a rectifying current. The input resistance data estimated a reversal potential (50 ± 1.5 mV) lower than the reversal potential elicited by the I_NSC (−38.8 ± 1.8 mV). These observations suggested the involvement of a potassium conductance, as using our solutions this ion has an estimated reversal potential of −84 mV. This hypothesis was tested using a voltage step protocol (in ACSF containing TTX) to first isolate both I_A and I_K and then determine the effects of IL-1β on these currents. Pulse protocols employing a 250 ms step from −70 to +30 mV (in 10 mV increments), with and without a 500 ms prepulse at −30 mV to inactivate I_A, were used to permit isolation of I_K and I_A by subtraction techniques (Washburn et al. 1999a; see Fig. 7). Bath application of 500 fM IL-1β caused a significant decrease in I_K to 84.4 ± 3.3 % of control in 4 of 7 (P < 0.005) DI_K cells (the IL-1β-responsive population), tested at the +30 mV step as illustrated in Fig. 7Bi in the absence of any change in conductance or slow capacitance. Although no recovery was seen from this decrease in current, this effect was unlikely to be the result of current run-down as application of 500 fM IL-1β had no significant effect on the subtracted I_A measured in the same group of cells, as shown in Fig. 7Bii. No significant change was seen in a small population of DI_K cells in response to 500 fM IL-1β in the presence of 100 pM IL-1ra (n = 3), with conductance and capacitance measurements again unchanged by drug application.

Figure 7. Voltage-clamp recordings displaying a decrease in I_K following IL-1β application.

Voltage-clamp step protocols (250 ms steps from −70 to +30 mV in the presence of TTX), with (as shown in A) or without a 500 ms prepulse at −30 mV to inactivate I_A, were used to permit isolation of I_K and I_A. A, the isolated I_K before (left) and after (right) 500 fM application of IL-1β, revealing a decrease in this current. Bi, overlaid control and IL-1β currents; C, I-V relationship of the 4 I_K cells that responded to 500 fM IL-1β (out of a total of 7 tested), illustrating that IL-1β caused a significant decrease in current in this subpopulation of SFO neurons in the +10 to +30 mV range (*P < 0.05). Bii, overlaid traces showing the complete lack of effect of IL-1β on the subtracted I_A.

DISCUSSION

The results of this study, showing that SFO neurons respond to IL-1β in concentrations (1 nM to 100 fM) that are similar to those seen in the periphery during an immune challenge (Ma et al. 2000), are in accordance with the hypothesis that the SFO may play a significant role in the induction of fever as a result of these IL-1β-sensing abilities. Our data demonstrate that femtomolar concentrations of IL-1β should be sufficient to trigger activation of SFO neurons, as seen by the depolarized membrane potential and increased firing frequency at these low concentrations. Increasing the dose of IL-1β to septic levels (1 nM; Ebong et al. 1999) did not produce the depolarizations seen at low doses but, rather, resulted in a sustained hyperpolarization, suggesting that such concentrations of this peptide may exert additional controls over the excitability of SFO neurons. These divergent observations might be explained by the actions of IL-1β on different receptors, a hypothesis that gains support from our observation that while the depolarizing effects of low doses of IL-1β on SFO neurons were abolished by IL-1ra, hyperpolarizations in response to high doses were still observed in the presence of this receptor antagonist. These data suggest that the hyperpolarizing effects may be the result of IL-1β interactions with a different receptor, such as the toll-like receptor 4 (TLR4), which in addition to showing some homology with the IL-1β receptor has been identified in the SFO (Laflamme et al. 1999). Further study will be necessary to clarify this issue.

Having established that the IL-1β-mediated depolarization was due to interactions with IL-1β receptors, we investigated the intrinsic conductances responsible for this response. Our voltage-clamp studies suggest that depolarization occurs primarily as a result of an increase in current produced by the opening of NSCCs. Previous work in our laboratory has identified a NSC conductance in SFO neurons that displays a reversal potential similar to that of the I_NSC evoked by IL-1β (Washburn et al. 1999b). In the present study, estimates of the theoretical change in membrane potential following IL-1β application, at a resting potential of −50 mV, where mean IL-1β-induced current is approximately 5 pA (see Fig. 6B), suggest a 5 mV depolarization in a cell with an input resistance of 1 GΩ, and a 10 mV depolarization in a cell with an input resistance of 2 GΩ, values that are in accordance with those recorded in response to IL-1β application in current-clamp mode (approximately 5–10 mV). These results demonstrate that activation of the seemingly small I_NSC can have significant effects on the membrane potential of SFO neurons.

Interestingly, beyond the −20 mV range, the difference current was no longer linear, indicating the influence of a voltage-dependent, rectifying current. Our input resistance data and the relatively high voltage activation of this current suggested the involvement of a voltage-gated potassium conductance. Subtraction protocols through which we have previously identified both I_A and I_K (Washburn et al. 1999a) revealed that IL-1β caused a significant decrease in I_K. A decrease in this outward current hinders the repolarization of the neuron during action potential recovery. This results in a slower return to baseline and a smaller afterhyperpolarization, leading to an increase in membrane potential and firing frequency. The decrease in I_K was longer lasting (no recovery observed in the time course of our recordings) than the effects on the NSCC, indicating the activation of separate signal transduction pathways, which have not been investigated in the current study, although other reports of IL-1RI signal transduction involving activation of NF-kappaB or mitogen-activated protein (MAP) kinases (DiDonato et al. 1997) suggest such pathways to be worthy of further study.

In this study, we have shown that these direct effects of IL-1β were primarily observed in the DI_K subset of SFO neurons while DI_A cells were for the most part unaffected. We have previously demonstrated that the latter population of SFO neurons project primarily to the hypothalamic PVN, but unfortunately have no direct evidence as to the projection sites of DI_K cells (Anderson et al. 2001). It has, however, been demonstrated that in addition to projections to neurosecretory cells of the PVN and supraoptic nucleus (SON), the SFO sends efferent projections to median and medial preoptic nuclei, the organum vasculosum of the lamina terminalis (OVLT) and the medial septum (MS) (Miselis, 1981; Lind et al. 1982), sites known to be intimately involved in the febrile response (Elmquist et al. 1997; Boulant, 2000; Matsunaga et al. 2001). The localization of thermosensitive neurons in the preoptic area (POA) (Boulant, 2000) suggests a mechanism whereby synaptic input from SFO neurons to this region could contribute to the development of fever as a consequence of modulation of the excitability of these thermosensitive neurons. An alternative possibility is that the IL-1β-sensitive population of SFO neurons project to the OVLT, a second CVO suggested to be involved in the febrile response (Blatteis et al. 1983; Stitt, 1990). However, the study by Takahashi et al. (1997) showing that lesion of the OVLT did not affect the febrile response induced by I.V. LPS implies that SFO neurons that innervate this area are not essential for the initiation of fever. A further possible projection site for IL-1β-sensitive SFO neurons is the MS, which contains projections to both the POA and PVN (Lind et al. 1982), with outputs from this region being responsible for co-ordinated activation of thermoregulatory, endocrine and autonomic centres essential to the development of fever.

In summary, we have demonstrated that a subpopulation of isolated SFO neurons respond to IL-1β with depolarization and an increase in firing frequency, effects that are the result of actions of this peptide at the IL-1β receptor. Combined activation of a NSCC and inhibition of I_K probably underlie these effects of IL-1β. These responses were observed at IL-1β concentrations that correspond with those produced at the onset of the immune response and are consistent with the idea that circulating IL-1β activates the SFO as an early and prerequisite component of the integrated febrile response to an immune challenge.