This study performed in conscious animals has demonstrated for the first time that during the development of the systemic inflammatory response, extracellular concentration of the purine nucleotide ATP increases in the anterior hypothalamus and this has a profound effect on the profile of the accompanying thermoregulatory febrile response. This release of ATP coincides with the development of the initial phase of fever and appears to be also followed by increases in hypothalamic adenosine and then inosine concentrations. ATP receptor antagonists administered to the site of ATP release all markedly augmented and prolonged the febrile response. This finding suggests that ATP, released in the anterior hypothalamus during systemic inflammation, is acting locally as an endogenous antipyretic mediator limiting the magnitude and the duration of the febrile response. Notably, ATP release was not seen in the posterior hypothalamus.
This is the first in vivo study in conscious animals to utilize enzyme-based biosensors to monitor in real-time the release of ATP and adenosine from the structures located deep in the brain. In our earlier studies in anaesthetized rats, we used biosensors to determine in various physiological conditions changes in ATP and adenosine concentrations in the structures located close to the exposed surface of the brain (Dale et al. 2002; Gourine et al. 2002a, 2005a,b). Here we have demonstrated that purine biosensors can be used for long-term measurements of ATP and adenosine levels in the deep brain structures in unanaesthetized experimental animals.
The ATP biosensor responds immediately to changes in ATP concentration and has a 10–90% response time of < 10 s (Llaudet et al. 2005). Thus, the sensors will quickly and accurately reflect the dynamic of changes in ATP concentration in the vicinity of the sensor. We found that during systemic inflammation the peak increase in anterior hypothalamic ATP concentration reaches ∼4 μm. This increase represents a dramatic elevation of extracellular ATP level over its basal concentration in the brain, which is estimated to be in the low nanomolar range (Phillis & O'Regan, 2002; Melani et al. 2005). Release of ATP coincides with the beginning of fever starting ∼18 min and reaching its peak ∼45 min after LPS injection. This increase in ATP concentration was found to be relatively transient, decreasing back to baseline within 120 min after LPS administration.
Although ATP biosensors require glycerol to operate, their responses are insensitive to variations in glycerol concentration over the range of 0.5–5 mm (Llaudet et al. 2005). To provide sensors with sufficient glycerol, recording sites were preinjected and the guide cannula was filled with glycerol-enriched aCSF. Although, ATP sensors retained > 70% of their initial sensitivity at the end of the experiment, the rate of glycerol washout from the recording site is unknown. If glycerol were to be rapidly diluted and washed away then the time to peak and the amount of ATP released may be significantly underestimated. However, this is unlikely because the enzyme–polymer layer of the biosensor has the ability to entrap and retain glycerol in amounts sufficient for the sensor to operate. Controls in which ATP biosensors soaked in 5 mm glycerol solution and then washed and repeatedly tested with ATP in glycerol-free media demonstrated that these biosensors retain their full sensitivity to ATP for up to 4 h (N. Dale, unpublished observations). These observations suggest that the time-course and the amount of ATP measured in our experiments represent quite accurately the dynamics of ATP release in the anterior hypothalamus.
In addition, if ATP were to be rapidly broken down to adenosine then measurements of the extracellular adenosine may also reflect the time-course of changes in ATP concentration. However, significant release of adenosine in the anterior hypothalamus during systemic inflammation was observed in only 3 out of 6 animals tested. It is not entirely clear why the remaining three animals showed no changes in adenosine concentration. All adenosine sensors retained ∼40% of their initial sensitivity at the end of the experiment and histological analysis of the sensor placements showed that the recording sites in all six cases were within the same general area of the anterior hypothalamus. It is likely that a high baseline concentration of extracellular adenosine might prevent detection of adenosine released in relatively small amounts during ‘normal’ physiological activity. It was shown previously that adenosine levels in the rat striatum were markedly higher (∼20-fold) shortly after implantation of the microdialysis probe as compared with 24 h after the surgery (Pazzagli et al. 1993). It is possible that the amount of cellular damage produced during sensor placement varied markedly between animals. Therefore, a higher baseline concentration of adenosine produced through greater tissue damage may be responsible for our inability to detect its release in some of the animals. Since adenosine sensors were gradually losing their sensitivity after placements into the hypothalamus (∼60% in 4 h) it was impossible to preimplant the sensors and conduct the experiments on the next day.
However, in the three animals that displayed adenosine release in response to LPS, the peak increase in concentration and time-course of release were remarkably consistent. The increase in adenosine concentration occurred some 55 min and peaked ∼3 h after LPS challenge. Thus, the increase in adenosine concentration commenced some 37 min after the onset of ATP release. As ATP is usually broken down to adenosine quite rapidly, this difference in time-course suggests that the production of adenosine during systemic inflammation is unlikely to originate entirely from prior release of ATP. It is not entirely surprising considering the findings by Frenguelli et al. (2007) who demonstrated independent and distinct mechanisms underlying release of ATP and adenosine during brain ischaemia. On the other hand, changes in inosine concentration closely followed changes in adenosine levels, strongly suggesting that released adenosine is rapidly converted to inosine.
Functional implications: ATP release in the anterior hypothalamus and its role in the febrile response during systemic inflammation
Fever is one of the most significant symptoms of sickness. It is induced and orchestrated by the central nervous system, specifically by the preoptic area/anterior hypothalamus. Pro-inflammatory cytokines such as IL-1β and others induce the febrile response by their actions in the anterior hypothalamus (Klir et al. 1994; Kluger et al. 1995; Gourine et al. 1998). Considering the vast amount of recent literature which implicates extracellular ATP in triggering P2X7-mediated release of proinflammatory cytokines (see introduction) we expected that ATP produced in this part of the brain would be responsible for local cytokine production and therefore would play an important role in the development of the febrile response during systemic inflammation. This hypothesis has received further recent support from evidence demonstrating a marked up-regulation in P2X7 receptor expression in the brain following peripheral LPS challenge (Choi et al. 2007) and our recent observation that systemic blockade of P2X7 receptors attenuates febrile and cytokine responses evoked by LPS in rats (Gourine et al. 2005c).
However, our data in the present study do not support this hypothesis. Although, ATP is indeed released in the anterior hypothalamus during systemic inflammation, localized blockade of P2 and specifically P2X7 receptors (at the site of ATP release in the anterior hypothalamus) did not reduce the febrile response. Instead, all three ATP receptor antagonists tested (PPADS, BBG and oATP) markedly augmented and prolonged the febrile response. Activation of different P2X and P2Y receptor subtypes is known to depend upon extracellular concentration of ATP (McLarnon, 2005). In this study the peak increase in ATP concentration (∼4 μm) recorded by the biosensors was well below the levels of ATP generally required to activate P2X7 receptors (in excess of 1 mm) (Hide et al. 2000; McLarnon, 2005). We therefore conclude that ATP released in the anterior hypothalamus is unlikely to play any major role in triggering production of proinflammatory cytokines and, by extension, is unlikely to be involved in facilitating the development of fever.
Conversely, our data suggest a different, but equally important, role for ATP-mediated signalling in the anterior hypothalamus during the development of the febrile response. The significantly augmented and prolonged fevers that occurred when the action of ATP was blocked by antagonists suggest that ATP normally acts in the anterior hypothalamus to limit the magnitude and duration of the febrile response.
However, ATP release is relatively transient (it reaches a peak some 45 min following the LPS challenge and decreases back to baseline within 120 min). The effects of the P2 receptor antagonists administered into the anterior hypothalamus on Tb for the most part occurred much later than this. Only BBG had a significant early effect on the febrile response – its effect on Tb was evident at 75 min after LPS injection. PPADS and oATP had significant effects on the febrile response starting from 150 min and 240 min after induction of systemic inflammation, respectively.
Actions of ATP receptor antagonists in the posterior hypothalamus had very little effect on the LPS-induced febrile response. Only oATP (in a higher dose of 100 μg) resulted in a small potentiation of fever at the very late stages of the febrile response. This small effect of the higher dose might be either due to a diffusion of the antagonist away from the injection site and its action at the anterior hypothalamic structures, or due to some non-specific action of the higher dose.
These data suggest that transient release of ATP in the anterior hypothalamus during the initial phase of systemic inflammation triggers a longer lasting mechanism that subsequently limits the febrile response. Interestingly, there is evidence that ATP in the low micromolar range (similar to that detected in this study) may inhibit cytokine (including IL-1β) production by cultured microglia via its actions at metabotropic P2Y receptors (Ogata et al. 2003). Inhibition of cytokine production by ATP released in the anterior hypothalamus could therefore be responsible for limiting the febrile response. We believe that this is unlikely. Although, PPADS may inhibit certain P2Y receptors, the effects of either BBG or oATP on P2Y receptors have not been described. In addition, our preliminary studies in rats revealed no significant effect of PPADS on IL-1β expression in the hypothalamus during LPS-induced systemic inflammation in rats (A. V. Gourine, D. M. Poputnikov, R. Gerstberger & V. N. Gourine, unpublished observations).
Alternatively, transient ATP release could conceivably trigger activation of one or more of the hypothalamic endogenous antipyretic systems which include release of the known central fever-reducing substances such as glucocorticoids, arginine-vasopressin, α-melanocyte stimulating hormone and nitric oxide (for recent review see Roth, 2006). For example, the ability of ATP to induce hypothalamic release of vasopressin – one of the most potent endogenous antipyretics – is well documented (Kapoor & Sladek, 2000). There is also evidence from the study involving our laboratory of a widespread co-localization of neuronal NO synthase and P2X receptors in the hypothalamus (Yao et al. 2003). Conceivably, ATP may trigger NO production which has been shown to have an antipyretic central action in rabbits (Gourine, 1995; Riedel, 1997) as well as in other species (for a review see Steiner & Branco, 2003). If there is an extended time delay in the action of these systems following their activation via ATP release, this may explain the delayed and long lasting effect of P2 antagonists on the development of the febrile response. Distinct mechanisms underlying the first and second phases of fever (see for example Steiner et al. 2006) may also account for the delayed effect of ATP receptor antagonists on the late phase of the febrile response.
Our data cast doubt on a role of P2X7 receptors in mediating the hypothalamic effects of ATP on fever. BBG and oATP were chosen for this study as they efficiently antagonize P2X7 receptors (Ralevic & Burnstock, 1998; North, 2002). P2X7 receptors are significantly less sensitive to blockade by the generic P2 antagonist PPADS (IC50∼50 μm), therefore this compound was used for comparison. However, the effects of BBG and PPADS on Tb during fever were not radically different. The effect of oATP was slightly less profound but qualitatively similar to that of PPADS. However, BBG and oATP are not just P2X7 receptor antagonists. BBG is effective at the rat P2X2 and human P2X4 receptors while oATP in the micromolar range partially (60%) reduces currents at P2X1 and P2X2 receptors (for review see North, 2002). Interestingly, BBG and PPADS appear to be equally potent in blocking P2X2 currents (North, 2002), suggesting that P2X2 subunit-containing receptors may be the most likely candidates to mediate the effects of ATP on the febrile response. Indeed P2X2 receptor subunits (along with P2X4 and P2X6) are the most abundant ATP receptors expressed by CNS neurones (North, 2002; Khakh & North, 2006) and have a widespread distribution throughout the hypothalamus (Xiang et al. 1998; Loesch et al. 1999; Kanjhan et al. 1999; Yao et al. 2003).
Receptors for extracellular ATP – both ionotropic P2X and metabotropic P2Y – are widely expressed in the CNS both in neurones and glia. ATP has been found to modulate neuronal activity in many parts of the brain and to contribute to the central nervous control of many physiological functions. In this study performed in conscious rabbits we observed that during development of the systemic inflammation the extracellular concentration of ATP markedly increases in the anterior hypothalamus. When released, ATP acts locally to limit the magnitude and duration of the febrile response. These data demonstrate the importance of directly measuring neurotransmitter release, rather than inferring it from the effect of antagonists on a particular physiological process – this study demonstrated that during fever, ATP release occurs significantly before any effects of P2 antagonists on Tb are observed.
Fever is just one of several behavioural and autonomic adaptations that occur during the systemic inflammatory response. Others include a decrease in locomotor activity, sleepiness, malaise and loss of food appetite. It is intriguing to speculate about the degree to which locally released ATP and adenosine (specific to the preoptic areas/anterior hypothalamus) play a role in the development of these responses. For example, the cholinergic basal forebrain and the ventrolateral preoptic area – both involved in the control of sleep (Saper et al. 2001; Basheer et al. 2004) – are close to the areas of ATP and adenosine release. As adenosine can act at these sites to promote sleep (Basheer et al. 2004; Morairty et al. 2004), this may partially explain why systemic inflammation is often accompanied by an increased sleep drive. In addition, it will be important to establish the cellular origin of ATP and adenosine release (neurones or glial) as well as the mechanisms that cause and mediate their release.