Male C57BL/6 mice weighing 18–25 g were used for the experiments and purchased from DooYeol Biotech (Korea). The mice were housed 3–5 per cage at a temperature-controlled room (23 ± 1 °C, 12 h/12 h light/dark cycle with lights on at 08:00) and maintained with standard lab chow (pellet diet) and water ad libitum except when food was removed for deprivation experiments. Experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) at Seoul National University and were consistent with the International Association for the Study of Pain (IASP) guidelines.

All procedures were prepared as previously described.¹⁸ 20 µl of 1% formalin (formaldehyde solution, 36∼38%, Junsei) was intraplantarly (i.pl) injected into the right hind paw with a 0.3 ml insulin syringe. Following formalin injection, the animals were immediately placed in a test chamber and recorded using a video camera for 40 minutes. The time mice spent licking was measured during each 5 minutes by an observer who was blinded to the treatment. Formalin-induced pain behaviors during 0–10 minutes after formalin injection represented the first phase and during 10–40 minutes after formalin injection represented the second phase.

Both AM 6545 (peripherally restricted CB1R antagonist, Tocris) and SR 141716 (CB1R antagonist, Tocris) were diluted in 0.9% saline with 10% DMSO and 1% tween 80 and then sonicated. Either AM 6545 or SR 141716 was injected at a dose of 10 mg/kg (i.p.) in a volume of 10 ml/kg body weight 30 min before formalin injection. To evaluate the peripheral effects of SR 141716, formalin test was performed after i.pl. injection of SR 141716 (10 µg per mouse), which was diluted in 1% formalin (20 µl/0.9% saline).

Mice were sacrificed 1 h after formalin injection. Quantitative reverse transcription PCR was performed as previously described.¹⁹ The primer pairs for activating transcription factor 3 (ATF3) was CCAGGTCTCTGCCTCAGAAG, CCGATGGCAGAGGTGTTTAT. The primer pairs for c-Fos was GGTGAAGACCGTGTCAGGAG, CCTTCGGATTCTCCGTTTCTCT. The primer pairs for cnr1 was ACGGTGTTTGCCTTCTGTAGT, CTGTGTTATTGGCGTGCTTGT. The primer pairs for the synthesizing enzyme N-acyl phosphatidylethanolamine phospholipase D (NAPE-PLD) was GACGCTGATGGTGGA AATGGA, AGGTGGTCGTAGTGGTTGTGA. The primer pairs for GAPDH was ATGGTGAAGGTCGGTGTGAAC, CATGGTGGTGAAGACACCAGTAG. All experimental groups relative to the naïve control group were calculated by the ΔΔCT method with GAPDH as the reference gene.

All procedures were prepared as previously described.¹⁸ Animals were sacrificed 2 h after formalin injection. Sections were pre-blocked with 5% normal goat serum (NGS). Sections were incubated in 1st antibody (PC38, Calbiochem, USA; 1:1000) for 48 h at 4 °C, and incubated in biotinylated goat anti-rabbit (BA1000, Vector laboratories, USA; 1:400) for 2 h at RT. Sections were processed with ABC kit (PK-6100, Vectastain ABC kit, Vector Laboratories, USA), visualized with DAB kit (DAB substrate kit for peroxidase, Vector laboratories, USA). After dehydration steps, all sections were mounted on slide grass with hardening mounting medium (Sigma, Germany) and examined under the bright-field microscope (DM5000B, Leica, Germany)

For the quantification of c-Fos expression, we confirmed that most c-Fos expression induced by injection of 1% formalin (i.pl.) localized in Lumbar (L) 4–5 segment. 4–6 sections with the highest expression of c-Fos per animal were chosen, and lamina I-VI and lamina I-II (superficial dorsal horn) of L 4–5 segments were selected for analysis. The number of c-Fos positive neurons was counted blindly, and the mean value was used as representative counts. Using image J, the image of selected sections was converted to a greyscale, background subtracted, enhanced and sharpened. The intensity threshold was adjusted and then analyzed using the “analysis particles” function.

The methods for making in vivo preparation were similar to those in our previous studies.^20–22 Briefly, mice were anesthetized with urethane (1.2–1.5 g/kg, i.p.) and placed on a warm (∼40 °C) plate. A thoracolumbar laminectomy at T13-L2 levels was performed to expose the dorsal surface of the lumbar enlargement of the spinal cord at L3-L5 levels. The mouse was then placed in a stereotaxic apparatus (ST-7M-HT, Narishige, Tokyo, Japan). The dura matter was removed, and the pia-arachnoid membrane was cut for making a small window to insert a tungsten electrode with an impedance of 10 MΩ (FHC, Bowdoin, ME, USA) using a micromanipulator (MHW-4–1, Narishige) at a fixed angle. The electrode was placed into the spinal dorsal horn, and multiunit neuronal firings were amplified with a differential extracellular amplifier (EX1, Dagan, Minneapolis, MN, USA). The signal was bandpass-filtered at 300–3 kHz and sampled at 25 kHz. Recorded signals were spike-sorted with a software (Spike2 ver. 6, Cambridge Electronic Design, Cambridge, UK) as previously reported.^23,24 We detected unit firings if they had basically biphasic shapes, and the amplitude was three times higher than SD of the baseline. If the amplitude and shape of firings were almost the same (see inset in Figure 3(Cb)), we used the firings as a single unit. Formalin (20 µl, 1%, i.pl) was injected into the paw ipsilateral to the recording site. We continuously recorded more than 40 min after the formalin injection.

All procedures were prepared as previously described.^25,26 Mice were anesthetized by pentobarbital (i.p., 50 mg/kg) and placed in dorsal recumbency. A 1 cm skin incision was made immediately caudal to the xiphoid process. The liver was retracted with a saline dampened cotton swab, and gentle traction was applied to the esophagus by lifting the stomach out of the peritoneal cavity. Esophagogastric junction was visualized with the aid of a surgical microscope, and a vagal segment was isolated then excised bilaterally. The incision in the abdominal wall and skin were closed separately. In sham-operated mice, the vagus was exposed but not excised. A formalin test was performed at least 1 week after vagotomy.

Statistical analysis was performed using GraphPad Prism version 6.0 (GraphPad Software, USA). A comparison between the two groups was made using the unpaired Student's t-test. For multiple comparisons, data were analyzed using the one-way ANOVA or two-way ANOVA followed by the Tukey test. Detailed statistics for each experiment were shown in the figure legend. Data are presented as mean ± SEM. Differences with p < 0.05 were considered significant.

In a previous study, we showed that 24 h acute fasting suppressed formalin-induced spontaneous pain only in the second phase, which was reversed by systemic administration of SR 141716 (i.p., 10 mg/kg), a CB1R antagonist.⁶ To determine the specific role of peripheral CB1R in this study, we used AM 6545 (i.p., 10 mg/kg), which shows markedly reduced brain penetration whereas SR 141716 (i.p., 10 mg/kg) are detected high concentration in the brain at 1 h after treatment of AM 6545 and SR 141716.¹² Both AM 6545 (i.p., 10 mg/kg) and SR 141716 (i.p., 10 mg/kg) were treated 30 minutes before formalin injection (Figure 1(A)), and we compared the effect of AM 6545 and SR 141716 (Figure 1(B)). As compared with the vehicle-treated group, both AM 6545 and SR 141716 significantly reversed fasting-induced analgesia (Figure 1(B) and (C)). These results suggest that the peripheral CB1Rs contribute to fasting-induced analgesia.

To identify the role of peripheral CB1R in dorsal root ganglion (DRG), we administrated CB1R antagonist into the hind paw. In the previous study, i.pl injection of SR 141716 has been shown to significantly inhibit the analgesic effect of AEA, even at a dose of 100 ng.¹⁵ However, although we co-administrated much higher concentration of SR 141716 (i.pl., 10 µg) with 1% formalin, SR 141716 did not block the analgesic effect of fasting (Figure 2(A)). Furthermore, in the formalin model, there was no difference in mRNA expression of CB1R (encoded by the cnr1 gene) in DRG between the free-fed group and 24 h fasted group (Figure 2(Ba)). Even after 24 h fasting, the formalin-induced ATF3 expression was not different between the two groups (Figure 2(Bb)).

Next, we examined whether fasting suppresses formalin-induced pain signals at the spinal cord level. The protein expression of formalin-induced c-Fos, an important marker of nociceptive neuronal activation,²⁷ was not different between the free-fed group and 24 h fasted group (Figure 3(A)). The mRNA expression of c-Fos also did not differ between the two groups (Figure 3(B)). Furthermore, the mRNA expression of cnr1 and NAPE-PLD (the major enzyme responsible for AEA) at the lumbar spinal cord remained unchanged after 24 h fasting in the formalin-induced acute inflammatory pain model (Supplemental Figure 1). In the acetic acid-induced visceral pain model, 24 h fasting also did not affect the mRNA expression of c-Fos at the thoracolumbar spinal cord, whereas the mRNA expression of cnr1 and NAPE-PLD tended to increase slightly at thoracolumbar spinal cord (Supplemental Figure 2). We recorded in vivo spinal neuronal firings in response to cutaneous formalin injection. Formalin injection into the paw elicited neuronal firings in the spinal dorsal horn in free-fed group. As shown in Figure 3(Ca) and 3(Cb), the time course of the unit firing number showed a biphasic response. In 24 fasted group, formalin injection also showed a similar biphasic response, and the firing unit number in the 2^nd phase was not different between free-fed and 24 fasted groups (Figure 3(Cc)). That in 24 fasted groups treated with AM 6545 was not also different between free-fed and 24 fasted groups (Supplemental Figure 3).

Collectively, our results showed that the CB1R expressed in DRG and its central inputs to the spinal cord might not play a critical role in fasting-induced analgesia.

Peripheral CB1Rs are known to regulate feeding behavior via the vagus nerve.^11,28 Therefore, we investigated whether the activation of CB1R from the gastrointestinal (GI) system could modulate fasting-induced analgesia via the vagus nerve. A formalin test was performed at least 1 week after vagotomy (Figure 4(A)). In consistent with a previous study,²⁹ subdiaphragmatic vagotomy itself significantly suppressed formalin-induced pain behavior only in the second phase (Figure 4(Ba) and (Ca)). As compared to the sham group, fasting-induced analgesia did not differ after subdiaphragmatic vagotomy (Figure 4(Ba) and (Ca)). Following subdiaphragmatic vagotomy, AM 6545 tended to reverse the fasting-induced analgesia, although the effect was not significant (Figure 4(Bb) and (Cb)). On the other hand, SR 141716 significantly reversed fasting-induced analgesia (Figure 4(Bb) and (Cb)). These results suggest that the CB1Rs in the GI system, rather than sensory neurons, may contribute to fasting-induced analgesic effects by transmitting fasting signals to the brain which also expresses CB1Rs.