It is widely agreed that prostaglandins contribute to nociception facilitation through sensitization of primary afferent sensory neurons at the site of injury and spinal circuitry [3, 25–27]. The role of supraspinal PGE2 in pain facilitation remains less explored despite studies which have demonstrated the presence of PGE2 receptors in the hypothalamus, hippocampus and PAG [6, 7]. A particularly strong positivity to the polyclonal antibody against an amino-terminal portion of EP3 receptor and mRNA have been found within the PAG . The presence of EP1 receptor, whose role in pain facilitation has previously been established [12–17], has instead never been ascertained within the PAG. Previous evidence has suggested a facilitatory role of PGE2 within the PAG on pain transmission [10, 11]. One of our previous studies has shown that VL PAG EP1 receptor blockade inhibited pain responses or misoprostol-induced effect . In the present study, the expression of EP1 receptor in the VL PAG was ascertained by western blot and immunohistochemistry. Our results demonstrate that EP1 receptors were widely expressed within the VL PAG and co-localize with the vesicular GABA transporter (VGAT) positive profile. The GABAergic neural population constitutes ~ 50% of total neural elements (the majority are active tonic interneurons) of the PAG controlling its intrinsic activity which affects the antinociceptive descending circuitry function [28, 29]. EP1 receptor is associated with Ca2+ mobilization and neurotransmitter release [3, 18]. Consistent with a previous finding in the striatum suggesting that EP1 receptor can be expressed on either the presynaptic or postsynaptic membrane , EP1 receptor stimulation in our study could be associated with GABA release. An increase in GABA release within the VL PAG has already been shown for misoprostol, a non-specific prostaglandin analogue , which activates all EP receptors. In another study, sulprostone, an EP3 receptor selective agonist, and PGE2 inhibited miniature excitatory postsynaptic currents (mEPSCs), with no action on inhibitory postsynaptic currents (mIPSCs), demonstrating the role of EP3 receptor in decreasing glutamate release from presynaptic sites . Changes in excitatory and/or inhibitory neurotransmitters have a relevant meaning within the PAG, since its stimulation and/or depression is consistent with pain inhibition and/or facilitation. Moreover, despite the potential for persistent exposure to prostaglandins at the peripheral site of injury or within the CNS, few studies have been performed to assess the effect of prostaglandin receptor expression and mediated responses in chronic pain conditions . In neuropathic pain conditions induced by the SNI of the sciatic nerve, a reduced expression of EP1 receptor was observed within the VL PAG 7 days after the surgery. A significant decrease of EP receptor in chronic pain conditions has already been observed in the DRG, spinal cord dorsal horn and sensory neurons [32, 33]. EP receptor downregulation appeared to be PGE2-mediated, since it was abolished by intrathecal administration of ketoralac, an inhibitor of the COXs, which proved to be overexpressed in inflammatory pain conditions [34, 35].
Selective stimulation of EP1 receptor by intra-VL PAG microinjection of ONO-DI-004, a selective EP1 receptor agonist, as well as the stimulation of all EP receptors by PGE2, decreased the latency of the thermoceptive reaction. The facilitation of nociceptive responses by intra-VL PAG PGE2 is consistent with previous studies demonstrating that the direct application of PGE2 within the VL PAG was able to facilitate nociceptive responses in lightly anaesthetized rats [10, 11]. Here, as well as confirming this previous evidence using a different themoceptive test; the tail flick, we also show that selective EP1 stimulation by ONO-DI-004 is able to induce the facilitation of nociception. Moreover, this study shows that L335677, a selective EP1 receptor antagonist, which was unable to alter the thermal threshold at a low dose, completely prevented the effects of intra-VL PAG microinjections of ONO-DI-004 and PGE2. The fact that EP1 selective antagonist completely prevents the PGE2-induced effect suggests that the PGE2-induced hyperalgesic effect requires the involvement of EP1 receptor. This dependence of large spectrum EP receptor agonist on each EP receptor has already been emerged with misoprostol, a synthetic PGE2 analogue, whose effect has been shown to require EP1, EP2, EP3 and EP4 receptor stimulation .
Moreover EP1 receptor blockade by L335677 (at higher doses than those used for antagonizing ONO-DI-004 and PGE2 effects) increased per se the latency of the thermoceptive reaction. Since EP1 receptor blockade by L335677 produced antinociception, it appears that this receptor is under tonic activation in facilitating the nociceptive response within the VL PAG. Therefore taken together, this study and previous ones [10, 11] suggest that apart from performing an inhibitory role, VL PAG also mediates facilitatory pain control. A tonic facilitatory role of EP1, EP2, EP3 and EP4 receptors, whose blockade determined analgesia, has already been found in formalin-induced persistent pain conditions . However, it has scarcely been investigated in neuropathic pain conditions . ONO-DI-004 and L335667 were both effective in changing the nociceptive reaction in a facilitatory and inhibitory manner, respectively in neuropathic pain conditions in this study. In this study it is significant to observe that neuropathic pain, which leads to almost a 50% reduction in EP1 receptor expression, did not alter the effect of EP1 ligands in changing pain response in the VL PAG according to a previous study in the sensory neurons and spinal cord . Possible explanations for the maintained pain-related activity of EP1 receptor stimulation/blockade in the face of its downregulation may be due to: i) the presence of spare receptors whose reduction in expression does not affect their related response; ii) loss of receptor on a certain cell population not involved in controlling pain at this level or iii) a minimal threshold concentration of EP1 receptor-associated transduction enzyme despite the elevated number of EP1 receptors in the VL PAG in normal conditions. Further studies are necessary to clarify this issue.
Prostaglandin-induced pain facilitation involves the recruitment of the pain-responding neuron population in the RVM . The PAG modulatory effect on pain control involves RVM neuron activity. In this study, intra-VL PAG ONO-DI-004 modified both spontaneous and tail flick-related activities of ON and OFF cells. In particular, it increased the spontaneous activity of the "pronociceptive" ON cells and reduced the spontaneous activity of the "antinociceptive" OFF cells, as a facilitating nociception drug is expected to do . ONO-DI-004 also decreased the onset of the ON cell burst and OFF cell pause, together with an increase in the burst frequency and pause duration; all of which are effects that are critical for pain facilitation. The same effects on RVM cell activity were produced by PGE2, which stimulates all EP1-4 receptors. In our study we also observed that the effect of ONO-DI-004, and more intriguingly that of PGE2, were completely prevented by L335677, a selective EP1 receptor antagonist. PGE2 has a high affinity for all EP receptors , thus the complete blockade of PGE2-induced behavioural and electrophysiological effects by L335677 would suggest that: i) the participation of all EP receptors is necessary for the nociceptive facilitation as in "an in series" circuitry; ii) receptors other than EP1 and EP3 are not present within the VL PAG, a hypothesis which is however contradicted by a previous study  and iii) EP2, EP3 and EP4 receptors are not involved in controlling nociceptive and ON and OFF cell activity responses. The most likely possibility in our opinion would seem to be that the involvement of all EP receptors is required for the nociceptive facilitation of PGE2, since apart from preventing misoprostol-induced hyperalgesia  within the DL PAG, individual EP1, EP2, EP3 and EP4 receptor antagonists also proved to be analgesic in formalin-induced persistent pain . We cannot exclude, however, that in the VL PAG sub-region, in healthy animals and in a different rodent species (rats versus mice) EP2, EP3 and EP4 are not involved in controlling nociception or RVM cell activity, or that the higher doses used in one of our previous studies  have involved other receptor subtypes. Further studies with selective EP2, EP3 and EP4 receptor antagonists or knock out mice are therefore necessary in order to clarify this issue. However, what seems important in the current study is that intra-VL PAG microinjection of L335677 inhibits nociception and reduces the spontaneous and tail flick-related activity of the ON cells in the RVM. Moreover, L335677 produced an increase in spontaneous OFF cell activity as well as an increase in the onset of the OFF cell pause, together with a decrease in the duration of the pause. Such effects are consistent with behavioural analgesia and could be important in experimental pain models such as inflammatory and neuropathic pain.
Our data shows that 7 days after sciatic nerve insult, RVM cell activity changes in such a way that the ongoing ON cell activity increased and that of the OFF cell decreased. Apart from the ongoing activity, the tail flick stimulated activity also changed in nerve injured rats. The burst and onset of the burst of the ON cells increased and decreased, respectively. Consistently, the pause and onset of the pause of the OFF cell increased and decreased respectively in neuropathic rats. Thus it appears that ON and OFF cell activity in the RVM undergoes functional phenotypic change after SNI which leads to ON cell hyperactivity and OFF cell hypoactivity after neuropathic pain induction. Changes in the pain descending system contributing to chronic pain symptoms have been reported [37–40]. In particular an increased activity of ON cell, together with a depression of OFF cell activity 1 week after SNI have already been observed within the RVM . Moreover in this study, as well as decreasing tail flick latency EP1 receptor stimulation was still able to increase and decrease ON and OFF cell activities as far as the ongoing or the flick related activity was concerned. More interestingly, EP1 receptor blockade by L335677, at the same doses used in healthy animals, increased tail flick latency and modified the ongoing and tail flick-evoked activity of ON and OFF cells, consistently with behavioural analgesia. Evidence of effectiveness of a selective EP1 receptor antagonist on hyperalgesia and allodynia in neuropathic pain state has already been reported [17, 42] after systemic administration.
The first part of the study showed the location of EP1 receptor within VGAT positive cell population within the VL PAG. EP1 receptor stimulation is associated with changes in calcium concentration and neurotransmitter release, which would increase GABAergic tone, thereby inhibiting the antinociceptive descending pathway. This would generate behavioural pain facilitation through a GABAergic interneuron activation of RVM ON cells and a direct inhibition of RVM OFF cells. Conversely, the tonically active EP1 receptor blockade would generate the opposite effect and dis-inhibit the PAG descending pathway, generating antinociception. EP1 receptor blockade-induced nociception inhibition at PAG level is associated with ON cell inhibition and OFF cell activation (and with the opposite electrophysiological effect in the case of ONO-DI-004 and PGE2). As far as the RVM ON and OFF cell involvement on intra-VL PAG ONO-DI-004, PGE2 and L335677 effect is concerned, the current study only shows correlative and synchronous changes in ON and OFF cell activity and thermonociceptive responses. Nevertheless there is increasing evidence to suggest that the switching off of the ON cells, which are the nociceptive facilitating neurons to the dorsal horns [43, 44] and activation of OFF cell activity are critical events in the production of antinociception. Another issue of the study regards the doses of ONO-DI-004, L335677 and PGE2 chosen. As far as the PGE2 is concerned, the doses used in the current study match those used in another study using the same intra-PAG administration route . ONO-DI-004 and L335667 have never been microinjected within the PAG. Therefore given that most synthetic EP receptor compounds mantain selectivity for their target receptor subtype in low nM, or an even lower range, we microinjected doses (few pgs in 0.2 μl) which are within or even under the limit of selectivity. Moreover, we tested for both L335677 (0.1 pg) and ONO-DI-004 (0.001 pg) tenfold lower doses, which proved to be devoid of activity.