The novelty of this study is the correlation of electrophysiological effects of CRF in the amygdala with pain-like behaviors under normal conditions in the absence of injury or disease. We show that CRF administered to the CeLC, a brain area that plays a critical role in the emotional-affective component of pain [21, 22], enhances or triggers nocifensive and affective behaviors by increasing synaptic transmission and neuronal output through postsynaptic CRF1 receptor-mediated PKA activation.
CRF facilitated synaptic transmission at the PB-CeLC synapse that provides unfiltered nociceptive information from spinal cord and brainstem to the CeLC and undergoes plasticity in different pain models [23–30]. Increased transmission at the PB-CeLC synapse correlates with pain- and anxiety-like behaviors [21, 22]. Analysis of miniature EPSCs indicated a post- rather than presynaptic action of CRF. CRF also increased spike firing, suggesting increased neuronal output. We did not attempt to determine the ionic basis for this cellular effect. Our previous studies using CRF receptor antagonists in the arthritis pain model implicated Kv3-type potassium channels in the effects of endogenous activation of CRF1 receptors; Kv3 channel regulate firing rate through action potential repolarization . However, a number of other effects of CRF on electrophysiological properties of amygdala neurons have been described, including inhibition of the slow afterhyperpolarizing potential (AHP) following evoked repetitive firing, mixed effects on the medium AHP , and increase of R-type voltage-gated calcium channels . Importantly, non-accommodating Type A neurons recorded in this study only show a medium AHP whereas the slow AHP is characteristic of accommodating Type B neurons . Therefore, the mechanisms of any membrane effects of CRF are likely complex and warrant the detailed analysis in a separate study.
The effects of CRF could be blocked with a CRF1 but not CRF2 receptor antagonist, demonstrating the presence of functional CRF1 receptors in the CeLC under normal conditions. This is not trivial because CRF1 antagonists had no effect on their own under normal conditions in previous studies from our group [29, 37] and others . The lack of involvement of CRF2 receptors in CRF-induced synaptic facilitation may be due to the lower affinity of CRF for this receptor type [3, 53] or different expression levels of CRF1 and CRF2 receptors in the synaptic circuit studied here [54, 55]. CRF2 receptor mRNA expression is more restricted than that of CRF1, at least under normal conditions; highest levels of CRF2 receptor mRNA in the brain are found within the lateral septum, ventromedial hypothalamus and choroid plexus, while medial and posterior cortical nuclei of the amygdala show moderate expression levels [55, 56].
CRF effects were largely blocked by inhibition of PKA but not PKC. CRF receptors can couple to a number of G-proteins to activate a variety of intracellular signaling pathways, and PKA and PKC appear to play particular important roles . PKA is a critical contributor to pain-related plasticity of CeLC neurons in the arthritis model [28, 45]. A consequence of PKA activation is the NR1 subunit phosphorylation of NMDA receptors in the CeLC and increased NMDA receptor-mediated synaptic transmission in the arthritis pain model [27, 28]. The present study found that CRF can increase a latent NMDA component through CRF1 receptor-mediated PKA activation, suggesting that CRF can engage processes similar to those that generate synaptic plasticity in pain models such as arthritis.
Interestingly, synaptic plasticity in the CeLC in neuropathic pain does not depend on NMDA receptors  although NMDA receptor antagonists are effective in the CeA in reducing nocifensive and affective pain behaviors in a neuropathic pain model . CRF and CRF mRNA are increased in CeA neurons in neuropathic pain  but increasing endogenous CRF in the CeA with a CRF-binding protein inhibitor had mixed effects in neuropathic pain, facilitating nocifensive responses while attenuating emotional-affective behaviors . The data may suggest that NMDA and CRF receptors engage different elements of the intra-amygdala circuitry in neuropathic pain, acting for example on inhibitory systems that modulate CeA processing  and neuropathic pain responses  and can be engaged by CRF1  or CRF2  receptors. Here we focused on the modulation of excitatory transmission at the PB-CeLC that correlates positively with pain behaviors [21, 22] although this study does not rule out additional sites of action of CRF in the amygdala network that should be explored.
In our study, the electrophysiological effects of CRF correlated with behavioral consequences. CRF increased audible and ultrasonic vocalizations, which represent supraspinally organized nocifensive and affective responses to aversive stimuli , and decreased thresholds for spinal reflexes. The results are consistent with the concept that increased CeLC output, here induced by CRF, facilitates spinal and supraspinal behaviors. It remains to be determined if this is accomplished through descending facilitation or disinhibition . Non-accommodating Type A neurons in the CeA project to brainstem and forebrain areas involved in the expression of aversive behaviors and pain modulation, including the periaqueductal gray. These brainstem projections arise not only from medial but also lateral regions of the CeA and involve strong interconnections between CeLC and substantia innominata [21, 40, 60, 61]. Lateral CeA projection neurons contain a number of neuropeptides, including CRF, neurotensin and somatostatin, and the latero-capsular region is the major site of extrahypothalamic CRF expression [12, 33]. Direct brainstem projections from CeA can be glutamatergic  but CRF-containing CeA neurons also include a population of GABergic neurons [63, 64]. Therefore, CeLC output can activate descending facilitation or inhibit descending inhibition (dis-inhibition) to produce the behavioral effects of CRF observed in our study. The results of our experiments in which the HPA axis response was suppressed with dexamethasone pretreatment distinguish this neuromodulatory function of CRF from its role as a stress hormone.
The significance of our findings is that increasing CRF in the amygdala can trigger pain-like behaviors in normal animals and these behavioral effects correlate with increased neuronal activity in the CeLC. Pain arising from altered brain functions in the absence of tissue injury represents a clinically important concept that could explain pain or increased pain sensitivity in conditions of anxiety, depression, or addiction such as alcohol dependence [4, 9, 13, 65] that involve the CRF system in the amygdala. With regard to pain processing and pain modulation, exogenous application of CRF (this study) and endogenous release of CRF measured indirectly with a CRF1 receptor antagonist in our previous studies [29, 37] appear to have similar facilitatory effects.
Some technical aspects of our study deserve consideration. Choice of drugs at appropriate concentrations is critical for pharmacological validity. We used selective compounds at concentrations that are well established in the literature [7, 66–68] and in our own previous studies [29, 35, 37, 38, 45]. The fact that a selective CRF1 antagonist inhibited the synaptic and behavioral effects of CRF clearly implicates CRF1 receptors. The lack of effect of a CRF2 receptor antagonist was not due to an insufficient drug concentration because the same concentration produced facilitatory effects in a pain model in our previous study . Likewise, the PKC inhibitor was used at a concentration near the high end of the selective range to ensure the lack of effect on CRF functions was not due to an insufficient concentration. A lower concentration of this inhibitor has been shown to be effective in modulating G-protein coupled receptor function . Additional caveats need to be considered for drug application by microdialysis. While the concentration of the drug in the microdialysis fiber is known, the drug dose administered can only be estimated. Comparative data from our previous microdialysis and in vitro studies using CRF receptor compounds [29, 35, 37, 70] indicate that the tissue concentration is at least 100 times lower than in the microdialysis probe due to the concentration gradient across the dialysis membrane and diffusion in the tissue. Therefore, drugs were dissolved in ACSF at a concentration 100 times that predicted to be needed. Microdialysis was chosen for drug delivery because it offers several advantages, including continued drug delivery and steady state levels without a volume effect . Differential effects of CRF1 and CRF2 receptor antagonists and PKA and PKC inhibitors here and in our previous studies [29, 35, 37, 38, 45] argue against non-selective drug effects at the concentrations used. Placement control experiments suggest that the drugs did not spread beyond a distance of 1 mm around the tip of the microdialysis probe, which is consistent with our previous estimates [35, 37, 38, 70]. The distance between the tips of the microdialysis probes in the CeLC (effective drug administration site) and striatum (ineffective control site) is about 2 mm. The striatum was selected as in our previous studies [43, 45, 46, 48] because it is located adjacent (dorsolateral) to the CeLC but does not project directly to the CeLC . Desensitization of CRF effects needs to be considered. However, the persistence of CRF effects in the presence of a CRF2 antagonist (Figure 1E) and PKC inhibitor (Figure 4E) and during prolonged drug application (Figure 4C) argues against the loss of effectiveness, e.g., due to desensitization, as an explanation for the inhibitory effect of the CRF1 antagonist and PKA inhibitor.