Figure 2 shows acquisition data for fentanyl (0.01 mg/kg/inj), butyrylfentanyl (0.03 mg/kg/inj), cyclopropylfentanyl (0.01 mg/kg/inj), and acetylfentanyl (0.1 mg/kg/inj) in male and female rats. Active and inactive responses are shown, while timeout responses are not. Fentanyl and all three FAs produced acquisition of self-administration, and the survival analysis did not reveal any sex differences in the rate of acquisition of self-administration for fentanyl (χ2 = 0.4778, d.f.=1, p = 0.4894), acetylfentanyl (χ2 = 1.319, d.f.=1, p = 0.2507), butyrylfentanyl (χ2 = 0.1097, d.f.=1, p = 0.7405), or cyclopropylfentanyl (χ2 = 0.2928, d.f.=1, p = 0.5884). Since the survival analysis demonstrated no significant effect of sex for any of the treatment groups, we combined the male and female acquisition data for each drug and plotted these data, as depicted in Supplementary Figure S1.

Acquisition of self-administration for fentanyl, FAs, and saline. Active (circles) and inactive (triangles) responses are shown for males (black) and females (red), with filled circles indicating significant differences from inactive responses on a given day (Šídák’s test, p < 0.05). Group numbers are fentanyl (male n = 13, female n = 11), acetylfentanyl (male n = 7, female n = 5), butyrylfentanyl (male n = 7, female n = 4–5), cyclopropylfentanyl (male n = 8–9, female n = 7), saline (male n = 5, female n = 5)

A two-way ANOVA for the acquisition data in male subjects revealed significant main effects of active vs. inactive responding for fentanyl (F[1, 234] = 64.03, p < 0.0001), butyrylfentanyl (F[1, 118] = 194.0, p < 0.001), cyclopropylfentanyl (F[1, 152] = 63.35, p < 0.0001), acetylfentanyl (F[1, 117] = 63.15, p < 0.001), and saline (F[1, 78] = 12.56, p < 0.001). The analysis also demonstrated a significant interaction of response x treatment day for fentanyl (F[9, 234] = 2.53, p < 0.0086), butyrylfentanyl (F[9, 118] = 5.22, p < 0.0001), and cyclopropylfentanyl (F[9, 152] = 3.33, p < 0.0009), but not for acetylfentanyl (F[9, 117] = 1.95, p = 0.0515) or saline (F[9, 78] = 0.544, p = 0.8377). Šídák’s post-hoc test showed that active responses in male rats were significantly greater than inactive responses on the last few days of training for all drug treatments but not for saline (see filled black circles in Fig. 2).

A two-way ANOVA for the acquisition data in female subjects revealed significant main effects of active vs. inactive responding for fentanyl (F[1, 198] = 11.04, p < 0.0011), acetylfentanyl (F[1, 80] = 88.88, p < 0.0001), butyrylfentanyl (F[1, 74] = 117.9, p < 0.0001), cyclopropylfentanyl (F[1, 118] = 35.28, p < 0.0001), and saline (F[1, 78] = 4.903, p < 0.0297). The analysis also demonstrated a significant interaction of response x treatment day for acetylfentanyl (F[9, 80] = 2.379, p < 0.0193), butyrylfentanyl (F[9, 74] = 4.037, p < 0.0003), cyclopropylfentanyl (F[9, 74] = 4.037, p < 0.0003), but not for fentanyl (F[9, 198] = 1.087, p = 0.3740) or saline groups (F[9, 78] = 0.9462, p = 0.4907). Šídák’s post-hoc test showed that active responses for female rats were significantly greater than inactive responses on the last few days of training for all drug treatments, with the exception of fentanyl (see filled red circles in Fig. 2). In the case of fentanyl acquisition in female rats, an elevated level of inactive nose poke responding by certain subjects on days 8, 9, and 10 eliminated significant differences between active vs. inactive responding on those days.

Following acquisition, animals continued to dose-effect testing. In Fig. 3, dose-response data for the average number of active responses in the last 2 days of testing are plotted, along with the average responses for the last 2 days of saline training. Fentanyl and all three tested FAs produced partial inverted U-shaped dose-effect functions, most likely due to the absence of doses on the ascending limb of the dose-response curve for each drug. For all drugs tested, a significant main effect of dose was observed (fentanyl F[4, 75] = 6.864, p < 0.0001; acetylfentanyl F[4, 44] = 12.33, p < 0.0001; butyrylfentanyl F[4, 40] = 5.816, p < 0.0009; and cyclopropylfentanyl F[4, 50] = 9.454, p < 0.0001), but there was no significant main effect of sex, nor a dose x sex interaction. Since we found no effect of sex on the dose-response functions, we combined the male and female data and re-plotted the dose-effect results, as depicted in Supplementary Figure S2. When the data are combined in this manner, the evidence for inverted U-shaped dose-response curves is better illustrated for fentanyl, butyrylfentanyl, and cyclopropylfentanyl.

Dose-effect curves for fentanyl and FAs compared to the last two days of saline. Filled symbols indicate significant differences from saline (Šídák’s test, p < 0.05). Active responses shown for males (black) and females (red), with filled circles indicating a significant difference from saline responses. Group numbers are fentanyl (male n = 5–13, female n = 8), acetylfentanyl (male n = 5–7, female n = 4–5), butyrylfentanyl (male n = 5–7, female n = 4), cyclopropylfentanyl (male n = 5–9, female n = 5–6)

Post-hoc analyses for the dose-response data revealed some subtle sex-specific effects, with significant differences from saline within each sex noted by filled symbols in Fig. 3. For fentanyl responding, males showed a significant difference from saline at 0.0003 mg/kg/inf (p = 0.0291) and 0.001 mg/kg/inf (p = 0.0092), while females showed a significant difference from saline only at 0.001 mg/kg/inf (p = 0.0113). For acetylfentanyl, males showed significant differences from saline at 0.003 mg/kg/inf (p = 0.0006) and 0.01 mg/kg/inf (p = 0.0004), and females similarly showed significant differences at 0.003 mg/kg/inf (p = 0.0320) and 0.01 mg/kg/inf (p = 0.0034). For butyrylfentanyl, males showed a significant difference from saline at 0.003 mg/kg/inf (p = 0.0156), whereas females did not show a significant difference from saline at any dose, perhaps due to higher variability in response rates for this drug. For cyclopropylfentanyl, males showed significant differences from saline at 0.0003 mg/kg/inf (p = 0.0252), 0.001 mg/kg/inf (p = 0.0139), and 0.003 mg/kg/inf (p = 0.0248), and females showed significant differences at 0.0003 mg/kg/inf (p = 0.0268), 0.001 mg/kg/inf (p = 0.0016), and 0.003 mg/kg/inf (p = 0.0145).

Following dose-effect testing, animals continued on to extinction training, receiving saline infusions for active responses (Fig. 4). The survival analysis revealed no sex differences in the rate of extinction of self-administration for any drug (fentanyl χ2 = 1.038, d.f.=1, p = 0.3082; acetylfentanyl χ2 = 0.7786, d.f.=1, p = 0.3776; cyclopropylfentanyl χ2 = 0.1566, d.f.=1, p = 0.6923). Extinction testing for butyrylfentanyl featured only a single non-censored male animal; thus, survival analysis was not conducted for this drug. While cyclopropylfentanyl appeared to support higher extinction responding for both sexes, further analysis of total extinction active responses revealed no significant effects for any drugs. Specifically, two-way ANOVA demonstrated no effect of drug (F[3,27] = 2.386, p = 0.0911), no effect of sex (F[1, 27] = 0.5764, p = 0.4543) and no sex x drug interaction (F[3,27] = 0.08443, p = 0.9680). Since survival analysis revealed no significant effect of sex on extinction, we replotted the combined extinction data, as depicted in Supplementary Figure S3.

Total responses across 10 days of extinction testing for fentanyl (male n = 7, female n = 3), acetylfentanyl (male n = 6, female n = 4), butyrylfentanyl (male n = 3, female n = 4), cyclopropylfentanyl (male n = 4, female n = 5)