Experimental design.

(A) Calibration day (Day 1) with heat and pressure pain calibration and Functional Threshold Power test (FTP) (see Table S1). (B) Experimental days (Day 2 and Day 3) with cycling task (outside MR) and fMRI task (inside MR) with the only difference in the drug treatment administered (SAL or NLX).

Behavioural and fMRI results for successful pain induction.

(A) Heat pain ratings in the SAL condition for all stimulus intensities (VAS 30, 50, 70) showed a significant main effect of stimulus intensity (P < 2×10-16). The P-value reflects the significant main effect of stimulus intensity from the linear mixed effect model (LMER). (B-D) Significant activation for the parametric effect (Heat VAS 70 > 50 > 30) in the SAL condition in the (B) right antIns (MNIxyz: 36, 6, 14; T = 10.35, Pcorr-WB < 0.001), (C) right dpIns (MNIxyz: 39, -15, 18, T = 7.65, Pcorr-WB < 0.001) and (D) right MCC (MNIxyz: 6, 10, 39; T = 7.47, Pcorr-WB = 0.001). Displayed are the uncorrected activation maps (Puncorr < 0.001) for visualisation purposes. (E-G) Bars depict mean parameter estimates from the respective peak voxels for all stimulus intensities across participants whereas dots display subject-specific mean parameter estimates at the respective stimulus intensity. P-values were calculated using the LMER model for the fixed effect of stimulus intensity. * P < 0.05, **, P < 0.01, *** P < 0.001. Error bars depict SEM (N = 39).

Successful implementation of high (HI) and low-intensity (LI) exercise.

(A) Power output (Watt), (B) heart rate in beats per minute (bpm), and (C) rating of perceived exertion (RPE; BORG scale) during low (LI; green) and high intensity (HI; purple) cycling were all significantly different. P-values were calculated using a paired t-test (two-tailed, power: N = 38, hear rate: N = 34, BORG: N = 39). * P < 0.05, ** P < 0.01, *** P < 0.001.

Behavioural results for the effect of drug treatment on pain.

(A) Heat pain ratings revealed a significant interaction between drug and stimulus intensity (P = 0.01). Mean heat pain ratings were significantly higher under naloxone NLX treatment (orange) compared to placebo (SAL) (blue) across stimulus intensities. The P-value indicates a significant interaction effect of stimulus intensity and drug. (B) Differences in heat pain ratings between NLX and SAL condition (NLX – SAL) at each stimulus intensity revealed a significant main effect of stimulus intensity (P = 0.02). (C-D) Heat pain ratings at all stimulus intensities in both drug treatment conditions are significantly higher in NLX (orange) compared to SAL (blue) conditions for both sexes. (E) Differences in heat pain ratings between NLX and SAL condition (NLX – SAL) at each stimulus intensity for females (white) and males (dark grey) revealed a significant main effect of stimulus intensity (P = 0.003) and sex (P = 0.03) with an interaction showing a trend (P = 0.06). P-values were calculated using the LMER model for the interaction of stimulus intensity and drug. * P < 0.05, **, P < 0.01, *** P < 0.001. Error bars depict the SEM (N = 39).

Effect of NLX in the periaqueductal grey (PAG).

(A) Activation for the interaction of stimulus intensity and drug in the PAG (MNIxyz: -2, -24, -8, T = 5.17, Pcorr-SVC = 0.01) superimposed onto an MNI template brain. (B) Parameter estimates for SAL (blue) and NLX (orange) conditions for the respective peak voxel in the PAG (MNIxyz: -2, -24, -8). (C) Difference between parameter estimates of the peak voxel between NLX and SAL condition at each stimulus intensity. (D) Time course of BOLD responses for SAL (blue) and NLX (orange) during high pain (VAS 70) in this voxel. The shaded areas depict the SEM. The grey solid lines indicate the start and end of the painful stimulus. The shaded grey area displays the approximate time window for BOLD response taking into account a 5s delay of the hemodynamic response function.

Effect of exercise intensity and drug treatment on pain modulation.

(A) No significant main effect of exercise intensity on pain ratings in the SAL condition (P = 0.44, blue bars. The P- value was calculated using the LMER model with exercise intensity. In a separate LMER model for the interaction of exercise intensity and drug treatment, this interaction effect was not significant (P = 0.91) but a significant main effect of drug treatment (P = 0.005) was evident. Bars depict the average pain ratings in the SAL (blue) and NLX (orange) conditions in both exercise conditions and the dots represent the subject-specific average ratings across all stimulus intensities. Error bars depict the SEM (N = 39). (B-D) Regions of interest (ROIs) in the (B) RVM, (C) PAG, and (D) frontal midline (comprised of ACC and vmPFC). (E-G) Parameter estimates extracted from both exercise and treatment conditions for the respective ROIs showed no significant effect of exercise intensity in the SAL condition (Table S10 – S12) as well as no interaction of stimulus intensity with drug treatment (Tables S14 – S16). n.s. = not significant, * P < 0.05, ** P < 0.01, *** P < 0.001.

Fitness level on difference pain ratings (LI-HI exercise) and mFC activation.

(A) Subject- specific differences in heat pain ratings (dots) between low-intensity (LI) and high-intensity (HI) exercise conditions (LI – HI exercise pain ratings) and corresponding regression line pooled across all stimulus intensities in the SAL condition. Fitness level (FTP) showed a significant positive relation to heat pain ratings with a significant main effect of FTP (P = 0.02) on difference ratings. (B) Cortical activation for contrast: exercise high > exercise low with mean-centred covariate FTP (weight-corrected) in the right mFC (MNIxyz: 6, 45, 10; T = 4.59, Pcorr-SVC = 0.05) across all stimulus intensities in the SAL condition superimposed onto an MNI template brain. (C) Differences between parameter estimates of LI and HI exercise conditions (LI – HI exercise parameter estimates) from respective peak voxel, plotted for each subject as a function of fitness level (FTP). Regression lines are visualised and shaded areas represent the SEM (N = 39).

Exercise-induced pain modulation potentially depends on sex, fitness level, and drug treatment in mFC.

(A) Exercise-induced pain modulation in the SAL condition for males (red) and females (blue). Males showed larger hypoalgesic responses with increasing fitness levels as indicated by positive difference ratings in the SAL condition. Females showed no association between fitness levels and difference ratings. (B) No exercise-induced pain modulation after blocking µ-opioid receptors with NLX in males (red) and females (blue). (C) The activation pattern in the mFC (MNIxyz: 12, 64, 2; T = 4.78, Pcorr-SVC = 0.039) resulting from a two-sample t-test (two- tailed) between males and females for contrast interaction of exercise and drug with covariate FTP superimposed onto an MNI template brain. (D-E) The difference in parameter estimates from the peak voxel in the mFC in (D) SAL and (E) NLX condition for males (red) and females (blue). Each dot represents the difference in parameter estimates between LI and HI exercise conditions (LI – HI exercise) for each subject averaged across all stimulus intensities. Shaded areas represent the SEM (female: N = 21, male: N = 18).