Formalin-induced pain prolongs sub- to supra-second time estimation in rats

Animals

Thirty-two male Sprague–Dawley rats (weighing 230–250 g on arrival, purchased from Charles River, Beijing, China) were used in this study. The rats were allowed to adapt to the laboratory environment for at least 1 week before the experiment was conducted, and they were handled (by stroking them or letting them move freely on the experimenter’s arms for 5 min) once per day to familiarise them with manipulation by the experimenter. They were housed individually in separate cages with food and water available ad libitum; but since the rats were able to get water from the task, we removed the water bottle from the feeding cage the day before the formal experiment. When a rat did not perform the task on any day, it was allowed to drink freely from a standard 250-ml bottle for 5 min. In addition, we recorded the rats’ body weights every 3–4 days and gave them an additional 5 min access to water when a loss of >10 g from the last recording was noted. During the experimental period, the rats’ body weights were maintained between 270 and 360 g.

The room in which the animals were housed was temperature and humidity controlled (22  ± 2 °C, 50%  ± 10% humidity), with a 12/12-h light/dark cycle (lights on at 8:00 pm). All experiments were performed in accordance with the National Institutes of Health guidelines for the care and use of laboratory animals (ISBN: 0-309-05377-3, First Printing, 1996) and approved by the Institutional Review Board of the Institute of Psychology, Chinese Academy of Sciences. The approve number is H16036. All rats were alive upon completion of this research and were used in preliminary experiments for another study.

Experimental apparatus

Temporal discrimination training and temporal bisection testing of each rat were conducted in the same operant box (21 cm W × 32.5 cm L × 42.5 cm H) located in a soundproof chamber. The box was made of acrylic, had a strip floor and was positioned above a waste catch pan. Two retractable levers (ENV-112CMP; MED Associates Inc., St. Albans, VT, USA) were mounted symmetrically on one wall of the box, 9 cm above the floor. A liquid dispenser (ENV-201A) and water receptacle (2.5 cm above the floor) were spaced equally between the levers to provide water as the reinforcer. As the to-be-timed auditory cue, a pure-tone stimulus (2,900 Hz, 65 dB) was presented by a tone generator (ENV-223AM) mounted on the wall above the liquid dispenser. An illuminated infrared detective nose-poke hole (ENV-114BM) was mounted in the middle of the wall opposite the levers, 2.5 cm above the floor. An indicator lamp (ENV-221M) was located directly above the nose-poke hole. An exhaust fan was mounted on the soundproof chamber. The operant box was connected to a computer that recorded the rats’ experimental events. The experimenter furnished all output and input devices mounted in the box. The box was cleaned with water and medical-grade alcohol after each session.

Experimental procedure

The temporal bisection task was conducted with 16 rats each for the sub- to supra-second and supra-second ranges. After training, each rat was first submitted to three test sessions (one per day) to stabilise its performance. Then, each rat performed five test sessions on five consecutive days (one per day): the day before formalin injection (baseline), immediately after formalin injection, and 1–3 days after formalin injection. Before the day of injection, the rats were assigned to formalin and saline control groups (n = 8 each) according to their baseline performance. All experiments were performed during the rats’ dark cycle.

Temporal discrimination training

Temporal discrimination training was performed using a procedure modified according to previous research (Brasted et al., 1997; Callu et al., 2009; Deane et al., 2017). It consisted of lever-press training (1–2 days), forced-choice training (3 days), and free-choice response training (12 days). In each trial of a lever-press training session (1-h duration, no limit on trial number), an anchor-duration tone was presented, and only the corresponding lever was extended. The anchor-duration tone was the short or long tone used for conditioning (0.6 or 2.4 s for the sub- to supra-second range and 2 or 8 s for the supra-second range). For the forced-choice training trials (n = 100, equal numbers with long and short anchor durations), each tone was triggered by a rat’s nose-poke behaviour. Only the corresponding lever was extended following the termination of the tone. For the free-choice response training trials (n = 100), the tones were triggered by the rats’ nose-poke behaviours, and their termination was followed by the simultaneous extension of both levers. On the first 6 days of free-choice training, the levers were retracted only after a rat provided the correct response; on the next 6 days, the levers were retracted immediately after any response. Correct responses were followed by the provision of drips of water as a reinforcer. Non-response for 5 s resulted in lever retraction and no reward. The criterion for the successful completion of temporal discrimination training was the achievement of correct response rates ≥ 85% on 3 consecutive days. For all training and testing sessions, the rats were allowed to move freely in the operant box.

Temporal bisection testing

During the testing stage of the temporal bisection task, a series of to-be-timed tones of anchor and intermediate durations was presented, and the rats were free to press any of the levers. In the testing sessions, each trial began with the illumination of the indicator light. The to-be-timed tone was initiated by a nose-poke. At the end of the tone, the two levers were extended simultaneously. Lever retraction was prompted by any lever press or non-response for 5 s. The average intertrial intervals (measured from lever retraction to the beginning of the next trial) per rat and session were 20 s (range, 19–21 s) for the 0.6–2.4-s range and 17 s (range, 13–19 s) for the 2–8-s range. One testing session consisted of 140 trials (30 s/trial): 14 trials each with tones of five intermediate durations without reinforcement, and 35 trials each with the two anchor-duration tones with reinforcement. Typically, the rats obtained about 65 reinforcements per testing session. For the sub- to supra-second–range (0.6–2.4-s) experiment, the five intermediate durations were set at 0.756, 0.952, 1.2, 1.51 and 1.9 s. For the supra-second range (2–8-s) experiment, the intermediate durations were set at 2.51, 3.17, 4, 5.04 and 6.35 s. The sequence of these trials was random, but with the constraint of no more than three consecutive intermediate-duration trials.

Establishment of pain model

Acute inflammatory pain was induced by injection of 50 µl 1% formalin solution into the plantar surface of each rat’s hind paw. An equal-volume saline injection was used as the control. After injection, the rats were returned to the operant box to perform the bisection task, and their nociceptive behaviours (paw licking and paw lifting) during the 70-min test sessions were video recorded. The durations of paw lifting and licking, measured by the experimenter in 5-min epochs, and the cumulative duration of nociceptive behaviours in phases I (0–10 min) and II (20–60 min) were calculated for analysis.

Data analysis

Curve fitting

The proportion of “long” responses (P_L,calculated by dividing the number of responses to the “long” lever by the total number of responses in trials with the same tone duration) to each to-be-timed tone in each temporal bisection session was used to analyse time perception. Curve fitting and parameter calculation were performed using the Prism software (version 5; GraphPad Software Inc., San Diego, CA, USA). Following previous work (Deane et al., 2017; McClure, Saulsgiver & Wynne, 2005; Ward & Odum, 2007), the final fitting curve is a cumulative Gaussian distribution function: (1)${\displaystyle F\left(t\right)=a+\frac{b}{\sigma \sqrt{2\pi }}{\int }_{-\infty }^t\left[\exp -\left(\frac{{\left(t-\mu \right)}^2}{2{\sigma }^2}\right)\right]\mathrm{d}t,}$ where F (t) is the P_L when the duration is equal to a given sample (t), µ  is the mean, and σ is the standard deviation (representing the slope of the function). Parameters a represents the low asymptote and b represents the range of function. The mean (µ) is also the PSE, defined as a 50% chance that the animal will provide a “long” response (P_L = 50%). The PSE reflects the subjectively perceived length of time, with increases and decreases therein interpreted as under- and overestimation, respectively, of the duration (Kamada & Hata, 2018; Kamada & Hata, 2019; Rey et al., 2017). The Weber fraction, an index of temporal sensitivity, was calculated by dividing the standard deviation by the PSE (Crystal, Maxwell & Hohmann, 2003). A decrease in the Weber fraction indicates an increase in temporal sensitivity (Deane et al., 2017; McClure, Saulsgiver & Wynne, 2005; Ward & Odum, 2007).

The R² statistic was used to quantify the goodness of fit. For each testing session, an individual rat’s P_L was fitted with this model to obtain the PSE and Weber fraction (Figs. 1F and 1G, etc.); the average P_L of each group of rats was fitted with this model to obtain the best-fitting curve (Figs. 1D and 1E, etc.). When R² was <0.7 in curve fitting for any individual rat, the PSE and Weber fraction were considered to be inaccurate and were discarded from the analysis.

Statistical analysis

To assess the post-treatment effect on the PSE, we calculated the mean PSE and Weber fraction, and calculated the change in the PSE relative to baseline using the following formula: (2)${\displaystyle \text{Change in PSE}=\left[\left(\text{current-session PSE}-\text{baseline PSE}\right)∕\text{baseline PSE}\right]\times 100\text{\%}.}$

Changes in the Weber fraction in the post-injection sessions were calculated using the same formula. The Prism (version 5; GraphPad Software Inc.) and SPSS (version 25; IBM Corporation, Armonk, NY, USA) software packages were used for graph generation and statistical analyses, respectively. Two-way repeated measure analysis of variance (two-way RM ANOVA) followed by a simple main-effect analysis was conducted to analyse the following variables: nociceptive behaviours (group * time); P_Ls and response latency for each duration in the temporal bisection task [(group * duration) or (day * duration)]; PSE, Weber fraction and omitted trials at baseline and on the injection day (group * day). Independent-sample t test was used to compare the differences of the cumulative duration of paw licking and lifting, the daily correct response rates, and the omitted trials and the changes in PSE/Weber fraction on each post-injection day between the two groups. If these data do not conform to the assumptions of the statistical method, we will use the correction method. The data are expressed as means ± standard errors of the mean, and the significance level was set at α <0.05. Data from trials with no lever-press response were omitted from the analysis.

Effect of formalin-induced pain on temporal estimation in the sub- to supra- second range

As shown in Fig. 1A, formalin injection induced a typical biphasic pattern of nociceptive behaviours. Compared with the saline group, the formalin group exhibited significantly more such behaviours during the temporal bisection task [interaction of treatment and time: F (13, 182) = 8.624, p < 0.001; treatment effect: F (1, 14) = 28.500, p < 0.001; time effect: F (13, 182) = 10.844, p < 0.001; Fig. 1A]. Cumulative duration of paw licking and lifting after formalin injection were increased significantly in phases I [t (14) = −5.711, p < 0.001; Fig. 1B] and II [t (14) = −4.433, p < 0.001; Fig. 1C] compared with those after saline injection.

We compared the correct response rates to anchor-duration tones between two groups at baseline (0.6 s: saline 96.75% ± 1.38%, formalin 96.07% ± 1.20%; 2.4 s: saline 94.29% ± 2.96%, formalin 95.34% ± 3.10%) and on the injection day (0.6 s: saline 90.71% ± 3.23%, formalin 90.57% ± 3.19%; 2.4 s: saline 96.01% ± 2.21%, formalin 97.37% ± 1.05%), respectively. Independent-sample t tests revealed no difference in the correct response rates at baseline or on the injection day, indicating that formalin treatment did not affect the animals’ temporal discrimination ability.

The effect of formalin injection on P_Ls of temporal bisection task was presented in Figs. 1D and 1E. Two-way RM ANOVA revealed no significant interaction between duration and treatment affecting the P_Ls at baseline [F (6, 84) = 0.489, p = 0.815, Fig. 1D] and on the injection day [F (6, 84) = 2.486, p = 0.23, Fig. 1E]. Nevertheless, the significant duration effect showed that the rats could discriminate the durations of the to-be-timed tones (p < 0.001). For each group, the differences of P_Ls between baseline day and injection day were also compared by Two-way RM ANOVA. A significant interaction of the day and duration was found in the formalin group [F (6, 42) = 2.486, p = 0.038] and a simple main-effect analysis showed that P_Ls on the injection day were higher than these at baseline at 0.952 s (p = 0.030) and 1.2 s (p = 0.017). No such effect was observed in the saline group [F (6, 42) = 0.790, p = 0.583]. R² values for curve fitting ranged from 0.86 to 0.99 (average: 0.94) in the two groups at baseline and on the injection day, with no difference between groups.

As presented in Fig. 1F, two-way RM ANOVA revealed significant interaction of the treatment and day affecting PSE values [F (1, 14) = 6.405, p = 0.024]. A simple main-effect analysis showed that the PSE was lower in the formalin group than in the saline group on the injection day [F (1, 14) = 4.67, p = 0.049], but not at baseline [F (1, 14) = 0.02, p = 0.899]. The Weber fraction, however, did not change significantly [interaction of treatment and day: F (1, 14) = 0.017, p = 0.897, Fig. 1G]. Moreover, we found a negative correlation between the total duration of nociceptive behaviours and the PSE on the injection day in all rats (r = −0.505, p = 0.046, Pearson product-moment correlation; Fig. 1H).

We also compared the omitted trials at baseline (saline, 4.125 ± 3.199; formalin, 0.875 ± 0.372) and on the injection day (saline, 4.250 ± 2.296; formalin, 3.750 ± 0.862). Two-way RM ANOVA revealed that the treatment had no significant effect in the omitted trials [treatment effect: F (1, 14) = 0.400, p = 0.537; interaction of day and treatment: F (1, 14) = 3.264, p = 0.092]. The response latency (defined as the time between lever extension and pressing) did not differ between groups at baseline [treatment effect: F (1, 14) = 2.140, p = 0.166; interaction of duration and treatment: F (6, 84) = 1.814, p = 0.106, Fig. S1A] or on the injection day [treatment effect: F (1, 14) = 0.113, p = 0.741; interaction of duration and treatment: F (6, 84) = 0.693, p = 0.656; Fig. S1B].

Effect of formalin-induced pain on temporal estimation in the supra-second range

In this range, formalin injection also induced a typical biphasic pattern of nociceptive behaviours during the temporal bisection session. As shown in Fig. 2A, formalin injection significantly increased nociceptive behaviours compared with saline injection [treatment effect: F (1, 13) = 26.183, p <0.001; time effect: F (13,169) = 11.190, p < 0.001; interaction of treatment and time: F (13,169) = 7.363, p < 0.001]. The cumulative duration of paw licking and paw lifting was significantly greater in the formalin group than in the saline group in phases I [t (13) = −6.577, p < 0.001; Fig. 2B] and II [t (13) = −3.155, p = 0.019; Fig. 2C].

Correct response rates to anchor durations was also analysed. Independent-sample t tests revealed no significant difference at baseline (2 s: saline 99.28% ± 0.47%, formalin 100.00% ± 0.00%; 8 s: saline 97.99% ± 1.05%, formalin 94.67% ± 2.29%) or on the injection day (2 s: saline 99.62% ± 0.38%, formalin 95.81% ± 3.74%; 8 s: saline 94.52% ± 1.13%, formalin 97.53% ± 1.16%), indicating that formalin treatment did not influence the temporal discrimination ability within this temporal range.

The effect of formalin injection on P_Ls in the supra-second range was presented in Figs. 2D and 2E. Two-way RM ANOVA revealed no significant interaction of duration and treatment affecting the P_L at baseline [F (6, 78) = 0.405, p = 0.874] or on the injection day [F (6, 78) = 0.680, p = 0.666]. However, the treatment had a significant main effect on the injection-day P_L[F (1, 13) = 5.186, p = 0.040]; no such effect was observed at baseline [F (1, 13) = 0.202, p = 0.660]. The significant duration effect revealed that the rats discriminated the durations of the to-be-timed tones (p < 0.001). For each group, two-way RM ANOVA was also used to compare the differences of P_Ls between baseline day (Fig. 2D) and injection day (Fig. 2E). No significant interaction of the day and duration was found in the formalin group [F (6, 36) = 0.804, p = 0.573] or saline group [F (6, 42) = 0.335, p = 0.915]. R² values for curving fitting ranged from 0.81 to 0.99 (average: 0.95) in the two groups at baseline and on the injection day, with no difference between groups [data from one rat were excluded because the R² values from all tests were low (0.734 ± 0.168)].

As shown in Figs. 2F and 2G, no significant interaction of the treatment and day affecting the PSE [F (1, 13) = 0.772, p = 0.396] or the Weber fraction [F (1, 13) = 0.892, p = 0.379; Fig. 2G] was observed. In addition, we observed no correlation between the PSE and nociceptive behaviours of all rats in the supra-second range (r = −0.044, p = 0.877, Pearson product-moment correlation; Fig. 2H).

Two-way ANOVA revealed no significant difference between omitted trials at baseline (saline, 4.571 ± 1.45; formalin, 10.875 ± 3.966) and on the injection day [saline, 7.571 ± 2.203; formalin, 10.500 ± 3.375; treatment effect: F (1, 13) = 1.199, p = 0.293; interaction of treatment and day: F (1, 13) = 1.503, p = 0.242]. At baseline, response latency did not differ between groups [treatment: F (1, 13) = 0.848, p = 0.374; interaction of duration and treatment: F (6, 78) = 1.120, p = 0.358; Fig. S2A]. A significant interaction of duration and treatment affected response latency was observed on the injection day [F (6, 78) = 4.169, p = 0.001]; no such interaction was observed at baseline [F (6, 78) = 0.418, p = 0.865; Fig. S2B]. A simple main-effect analysis showed that the formalin treatment prolonged the response latency in the 4-s trial (p = 0.035).

Post-treatment effect of formalin-induced pain on temporal estimation

The correct response rates of temporal bisection tasks performed in the 0.6–2.4-s range on post-injection days were analysed. Independent-sample t tests revealed no significant difference in the correct response rates to anchor-duration tones on post-injection day 1 (0.6 s: saline 94.94% ± 1.99%, formalin 93.51% ± 3.42%; 2.4 s: saline 95.89% ± 2.22%, formalin 95.27% ± 1.32%), 2 (0.6 s: saline 95.63% ± 1.46%, formalin: 95.56% ± 2.61%; 2.4 s: saline 96.05% ± 0.93%, formalin 96.04% ± 1.70%), or 3 (0.6 s: saline 96.78% ± 1.66%, formalin 97.82% ± 1.44%; 2.4 s: saline 98.21% ± 0.93%, formalin 96.07% ± 1.20%). These results indicate that the post-injection effect of formalin treatment did not influence the animals’ temporal discrimination ability for the anchor durations.

The results of temporal bisection tasks performed in the 0.6–2.4-s range on post-injection days are presented in Figs. 3A–3C. No significant interaction of treatment and duration affecting the P_L was observed [post-injection day 1: F (6, 84) = 1.861, p = 0.097; post-injection day 2: F (6, 84) = 1.025, p = 0.415; post-injection day 3: F (6, 84) = 1.706, p = 0.130]. The main treatment effect also was not significant [post-injection day 1: F (1, 14) = 1.304, p = 0.273; post-injection day 2: F (1, 14) = 2.300, p = 0.152; post-injection day 3: F (1, 14) = 1.359, p = 0.263]. The main effect of duration was significant in all tests [post-injection day 1: F (6, 84) = 102.703, p <0.001; post-injection day 2: F (6, 84) = 92.340, p < 0.001; post-injection day 3: F (6, 84) = 116.652, p < 0.001], indicating that the rats could discriminate the durations of the to-be-timed auditory stimuli. R² values for curve fitting ranged from 0.87 to 1 (average: 0.96), with no significant difference between groups on any of the 3 days.

As shown in Figs. 4A and 4B, we compare the change in the PSE and Weber fraction in the 0.6–2.4-s range from baseline on each post-injection day (Original PSE, day 1: saline 1345.375 ± 53.468, formalin 1183.15 ± 111.639; day 2: saline 1413.375 ± 124.251, formalin 1140.388 ± 128.744; day 3: saline 1435.875 ± 82.933, formalin 1221.438 ± 107.257; Original Weber fraction, day 1: saline 0.217 ± 0.059, formalin 0.222 ± 0.054; day 2: saline 0.231 ± 0.063, formalin 0.254 ± 0.038; day 3: saline 0.231 ± 0.048, formalin 0.222 ± 0.032). Independent-sample t tests showed that the formalin treatment tended to reduce the change of PSE on post-injection days 2 [t (9.353) = 2.143, p = 0.060] and 3 [t (8.527) = 2.238, p = 0.054], see Fig. 4A.

Omitted trials from the three post-days (day 1: saline 6.375 ± 3.201, formalin 3.625 ± 1.992; day 2: saline 5.000 ± 3.082, formalin: 2.500 ± 0.810; day 3: saline 5.250 ± 3.521, formalin 1.625 ± 0.431) were also analysed. Independent-sample t tests revealed no post-effect of formalin treatment in the omitted trials from any day. Two-way RM ANOVA (duration * treatment) revealed that the treatment had no significant effect on response latency on any of the 3 days [treatment effect: F (1, 14) ≤ 1.924, p ≥ 0.187; interaction of duration and treatment: F (6, 84) ≤ 1.862, p ≥ 0.097, Fig. S3].

In the 2–8-s range, the correct response rates to anchor-duration tones did not differ significantly on post-injection day 1 (2 s: saline 100.00% ± 0.00%, formalin 97.96% ± 1.20%; 8 s: saline 94.57% ± 1.67%, formalin 95.45% ± 1.97%), 2 (2 s: saline 99.255 ± 0.49%, formalin 100.00% ± 0.00%; 8 s: saline 97.14% ± 2.33%, formalin 96.73% ± 2.01%), or 3 (2 s: saline 100.00% ± 0.00%, formalin 98.78% ± 0.85%; 8 s: saline 93.77% ± 2.17%, formalin 95.00% ± 2.27%). These results indicate that the formalin treatment did not influence the animals’ temporal discrimination ability for the anchor durations.

The results of temporal bisection tasks performed in the 2–8-s range on post-injection days are shown in Figs. 3D–3F. Two-way RM ANOVA revealed no significant interaction of treatment and duration affecting the P_L on any day [post-injection day 1: F (6, 78) = 0.303, p = 0.934; post-injection day 2: F (6, 78) = 0.700, p = 0.651; post-injection day 3: F (6, 78) = 0.695, p = 0.654]. The effect of treatment on P_L also was not significant [post-injection day 1: F (1, 13) = 0.097, p = 0.761; post-injection day 2: F (1, 13) = 0.265, p = 0.615; post-injection day 3: F (1, 13) = 0.331, p = 0.575]. Duration had a significant effect on the P_L in all tests (p <0.001). R² values for curve fitting ranged from 0.83 to 1 [average: 0.95; the PSE and Weber fraction from one rat were excluded because the R² values were low (average: 0.6) on post-injection day 2], with no significant difference between groups on any of the 3 days.

As shown in Figs. 4C and 4D, the treatment had no significant effect on the change of PSE or Weber fraction from baseline on any of these post-injection day (Original PSE, day 1: saline 3.427 ± 0.216, formalin 3.403 ± 0.224; day 2: saline 3.486 ± 0.116, formalin 3.784 ± 0.23; day 3: saline 3.500 ± 0.180, formalin 3.800 ± 0.419; Original Weber fraction , day 1: saline 0.191 ± 0.026, formalin 0.185 ± 0.043; day 2: saline 0.130 ± 0.029, formalin 0.207 ± 0.042; day 3: saline 0.186 ± 0.038, formalin 0.260 ± 0.056).

Omitted trials from the three post-injection days (day 1: saline 6.750 ± 4.259, formalin 13.143 ± 5.501; day 2: saline 7.875 ± 4.673, formalin 9.857 ± 4.752; day 3: saline 8.125 ± 4.525, formalin 11.000 ± 3.9) were analysed. Independent-sample t tests revealed no effect of the formalin treatment on omitted trials from any day. The response latency did not differ significantly on post-injection days 1 and 2 [treatment: F (1, 13) ≤ 2.58, p ≥ 0.132; interaction of duration and treatment: F (6, 78) ≤ 1.365, p ≥ 0.239; Fig. S4]. Interaction of treatment and duration affecting response latency was observed for post-injection day 3 [treatment: F (1, 14) = 0.913, p = 0.357; interaction of duration and treatment: F (6, 78) = 2.666, p = 0.021; Fig. S4C]. A simple main-effect analysis revealed significantly more response latency in the 4-s trials in the formalin group (p = 0.022).