In the study herein presented, we explained, at least in part, the possible mechanisms underlying the divergent susceptibility to acute inflammatory pain in mice selected for high (HA) and low (LA) swim-stress-induced analgesia (SSIA). Our results were in accordance with an early study by Lutfy and colleagues [31] who were the first to demonstrate that HA mice were more sensitive to inflammatory pain than LA mice in the formalin test. We observed a similar pattern in the present study, where an intraplantar injection of formalin produced more profound nocifensive behaviours in the HA than the LA mice in the tonic phase (Figure 1). Differential levels of microglial activation could serve as one possible mechanism, as minocycline more profoundly suppressed formalin-induced paw licking in the HA than in the LA mice (Figure 5). This theory was supported by higher mRNA levels of pro-inflammatory markers Aif1 and Il-1β in the HA mice as early as 15 min post-formalin, while their levels were either unchanged or decreased in the LA line (Table 6 and Table 7). The Aif1 gene encodes the Iba-1 protein that regulates actin cross-linking, which is essential for morphological changes in spinal microglia during the early stages of activation and its expression correlates with hypersensitivity. As Aif1 expression is also present in ramified microglia, Il-1β, a hallmark neuroinflammatory marker, was additionally studied [32,33,34].

We further enhanced the original study by Lutfy [31] by additionally measuring pain thresholds in the interphase of the formalin test. The interphase is characterized by the cessation of excessive neuronal firing provoked by an irritant, such as formalin. Henry et al. argued that this phase is governed by a yet unknown spinally mediated active inhibition mechanism [35]. Our current investigation clearly showed that the HA mice presented significantly less nocifensive behaviours in the interphase than the LA mice (Figure 1). Thus, we hypothesized that this active inhibitory mechanism could be attributed to the endocannabinoid system, as our previous studies indicated substantial differences in its activity between HA and LA mice [17,36]. The HA mice showed a significant increase in spinal and amygdalar Cnr1 transcripts (Table 1), while Cnr2 mRNA levels were elevated only in the amygdala (Table 2). Additionally, increased concentrations of anandamide (AEA) and 2-AG were detected in the spinal cord (Figure 4). Moreover, rimonabant—a selective Cnr1 receptor antagonist—dose-dependently intensified pain behaviour in the interphase, but only in the HA line (Figure 3).

Compelling evidence exists for a compensatory peripheral and central increase in AEA and 2-AG in many inflammatory states such as arthritis and osteoarthritis as well as in relapsing multiple sclerosis patients [37,38]. In addition, enhanced AEA signaling resulting from fatty-acid amide hydrolase (FAAH) loss of function was associated with pain insensitivity [3]. These findings imply a significant role of endocannabinoids in neuronal excitability and may account for the divergence in inflammatory pain sensitivity between the HA and LA mice seen in the interphase. Their effect is executed by either a direct action at the Cnr1 receptors and/or modulation of ion channel conductance. It has been recognized that endocannabinoids inhibit Ca²⁺ influx in a Cnr1-dependent and independent manner [39] and also inhibit the tetrodotoxin (TTX)-sensitive Na⁺ channels in cortical and hippocampal neurons [40,41]. Moreover, endocannabinoids activate inwardly rectifying K⁺ channels and induce neuron hyperpolarization [42].

It is commonly acknowledged that spinal Cnr1 receptors are important modulators of nociceptive transmission as their intrathecal delivery produces antinociception in numerous models of pain [43,44,45]. Namely, their activation reduces dorsal horn C- and Aδ-fiber electrical activity both in anesthetized animals and in animals with peripheral inflammation [46,47]. Moreover, loss of spinal Cnr1 receptors produces hyperalgesia [48]. Endocannabinoid binding to Cnr1 receptors in the amygdala is important for the activation of the descending inhibitory pain pathway. Namely, they enhance glutamatergic neurotransmission by suppressing GABA-ergic inhibitory interneurons and increase the firing of periaqueductal gray matter (PAG) projection neurons. This allows stronger input from the rostral ventromedial medulla (RVM) “off” cells to the spinal cord dorsal horn and triggers analgesia. This mechanism is responsible for the effect of WIN 55212-2 microinjection into the amygdalar basolateral nucleus, where it suppresses formalin-induced pain behaviour in the first and tonic phases [49]. Moreover, overexpression of Cnr1 and Cnr2 receptors in the amygdala could possibly serve as an adaptive anxiolytic response to a stressful stimulus such as irritant injection. For instance, it was demonstrated that mice overexpressing Cnr2 receptors were more resistant to stressful stimuli and that Cnr1 receptors mediate the anxiolytic effects of drugs [50,51]. Additionally, stimulation of both Cnr receptor subtypes was implicated in the early, first line mechanism impairing fear memory consolidation and generalization to facilitate aversive memory extinction [52].

To our surprise, nocifensive behaviours seen in the tonic phase in the HA line were not dependent on Cnr1 receptor activation as even the highest dose of rimonabant failed to intensify inflammatory pain in this line. It is possible that increased levels of AEA and 2-AG paired with increased receptor availability in HA mice could alleviate the pronociceptive effect of Cnr1 blockage. Another hypothesis assumes that Cnr1 receptor blockage triggers the compensatory activation of Cnr2 receptor-mediated signaling in the amygdala and in the spinal cord. As shown before, even low-grade inflammation is sufficient to induce anandamide synthesis and enhance spinal Cnr2 receptor expression [53]. The sole role of the Cnr2 receptors in pain sensitivity was evidenced before in nerve injury models, where Cnr2 knockouts showed enhanced ipsilateral pain sensitivity and also developed mirror hyperalgesia [54,55]. Despite no differences in post-formalin spinal Cnr2 expression in HA mice, these mostly microglia-residing receptors, could still be a possible target for spinal 2-AG binding to produce antihyperalgesia in conditions of Cnr1 inhibition, especially when HA mice showed more pronounced microglial activation reflected by an increase in Aif1 and Il-1β transcripts in response to formalin than LA mice. Moreover, AEA release could be an attempt to reduce microglial polarization towards the inflammatory M1 phenotype, which is Cnr2-receptor-dependent [21]. Additionally, mice from the LA line show impaired G-protein Cnr2 receptor coupled function [17], hence a more prevalent involvement of Cnr1 receptor signaling in the tonic phase and compensatory upregulation of Cnr2 receptors in the PAG and hypothalamus.

The contribution of the endogenous opioid system to inflammatory pain sensitivity in HA and LA mice was also addressed. However, naloxone was ineffective in both lines and both phases (Figure 2), thus providing evidence of the endogenous opioid system not playing a major role in modulation of formalin-induced acute inflammatory pain. One possible explanation for the compromised contribution of the endogenous opioid system in formalin-induced pain sensitivity is the downregulation of Oprd1 expression in the PAG of HA mice (Table 4). As we have previously shown, δ-opioid receptors significantly influenced nociceptive thresholds in HA mice. For instance, HA mice showed higher basal hypothalamic Oprd1 expression than their LA counterparts (Table 4). In addition, we have previously identified an A107V substitution in the Oprd1 gene that renders it less responsive to ligands [56]. An upregulation of the Oprk1 gene coding the κ-opioid receptor was also evidenced in the PAG of HA mice (Table 5). This phenomenon could be associated with the triggering of aversion circuits in response to a stressful stimulus such as formalin injection [57,58]. HA mice are more vulnerable to stress than LA mice, thus acute stress could induce Oprk1-mediated adaptive changes to trigger aversive and defensive behaviours. The ineffectiveness of naloxone in LA mice may be explained by the widely recognized opioid system hypoactivity and opioid ligand insensitivity in this line [36,59]. Additionally, as shown in the current study, formalin induced Oprm1 downregulation in the PAG in LA mice, which could also contribute to naloxone insensitivity (Table 3).