Animals

All experimental procedures were in accordance with the Dutch law on animal research in full agreement with the Directive 2010/63/EU and approved by the Netherlands Central Commission for Animal Experiments (Permit Number AVD11200202115388). Local approval and supervision were provided by the Animal Welfare Body at the Vrije Universiteit Amsterdam.

Twelve adult male Wistar rats (Rattus norvegicus domestica, 330 ± 34 gram prior to surgery) were used in this study. Only male rats were used because of the sex difference in pain response reported in literature (Cairns et al. 2001; Capra and Ro 2004). Upon arrival (9 weeks of age), rats were housed in pairs under a 12-h light/dark cycle. After EMG electrode implantation surgery (12 weeks of age), rats were housed in the same cage but separated by a cage divider, allowing to see and smell each other. Rats were allowed to move freely in the cage with access to food and water ad libitum.

Study preregistration

This study was preregistered at PreclinicalTrials.eu prior to conducting the research (registration number PCTE0000367). The preregistration adheres to the disclosure requirements of the institutional registry.

Experimental protocol

After one week of acclimatization, the rats were trained to run on a treadmill (Exer 3/6, Columbus Europe Instruments, Dublin, Ireland) for two weeks, then four pairs of Teflon-insulated fine-wire electromyography (EMG) electrodes (7SS-1T, Science Products, Hofheim, Germany) were implanted into the multifidus muscle (MF) and medial longissimus muscle (ML) between L4 and L5 vertebra bilaterally. After two weeks of recovery, in vivo measurements were performed to collect MF and ML EMG signals, as well as spine and pelvis kinematics, before and after hypertonic saline (5.8%) injection into the MF muscle (Fig. 1a).

Fig. 1

Overview of experimental protocol and data collection procedures. (a) Experimental timeline. * age of the rat.(b) in vivo measurement setup. (c) Illustration of motion tracking markers and the convention of joint angles. (d) (left) Representative band-pass filtered electromyograms (EMG) and (right) rectified (black line) and filtered (light grey line) EMG signals of bilateral multifidus (MF) and medial longissimus (ML) muscles from one stride cycle of the rat trotting on the treadmill at 0.5 m/s before hypertonic saline injection (rat S10). Bottom row shows the hind paw position of the rat during locomotion

Treadmill locomotion training

Prior to implantation of the EMG electrodes, rats were trained to run on a motor-driven treadmill for two weeks on a daily basis. Each training lasted for 10–20 min with increasing speed up to 0.6 m/s and consisted of several 2–3 min running sessions. Food reward followed by approximately 5 min rest was provided upon finishing each running session. To promote running, an electrical grid located behind the treadmill delivered a mild electrical stimulus (repetition rate: 2 Hz, stimulus current: max 1 mA, stimulus duration: 200 ms) once the animal stepped on the platform behind the treadmill. The electrical stimulus was automatically switched off once the rat stepped on the grid for a third time. In addition, before touching the grid, the rats would first come into contact with a small object hanging in front of the grid. This object served as a warning signal to avoid stepping on the platform.

Surgical procedures for EMG electrodes implantationAnimal Preparation

Carprofen (3 mg/kg, Rimadyl®, Zoetis B.V., Capelle a/d Ijssel, The Netherlands) was administrated subcutaneously 12 h before surgery. Both carprofen (3 mg/kg) and buprenorphine (0.02 mg/kg, Buprecare®, Ecuphar NV, Oostkamp, Belgium) were administrated subcutaneously 30–60 min before surgery. The rats were anaesthetized by isoflurane (induction in a box: 3–5%, maintenance via nose cone: 1–2%), then mounted in a stereotaxic frame (David Kopf Instruments, Tujunga, CA, USA) and placed on a heating pad. Eye ointment was applied during surgery to prevent dehydration. The local anesthetic Ropivacaine (2 mg/kg, Fresenius Kabi Norge AS, Halden, Norway) was applied several minutes before incision. Hind paw pain reflex, breathing rate, and rectal temperature were monitored throughout the surgery.

EMG connector

The EMG connector was mounted on the head of the animal. A skin incision (~ 1 cm) was made on top of the skull, and several drops of lidocaine with HCl (1%, 10 mg/ml, B. Braun, Melsungen, Germany) were locally applied to prevent excessive bleeding. Another skin incision (~ 1–2 cm) was made over the lower lumbar spine, then the electrode wires were threaded subcutaneously from the head down to the dorsal lumbar region. After exposing the skull, four stainless steel screws were placed on the skull, two in front of the bregma and two behind. The EMG connector was placed between the four screws, and anchored to the skull with dental cement (RelyX Unicem2 Automix, 3 M ESPE, Germany) to encapsulate the bottom part of the connector and all the screws. The skin was closed with sutures (4 − 0, ETHIBOND EXCEL, non-absorbable, ETHICON).

Implantation EMG electrodes

After securing the connector on the head, four pairs of EMG electrodes were implanted bilaterally into MF and ML between the L4 and L5 vertebral levels, using procedures described in our previous study (Bernabei et al. 2017) and Tysseling et al. (Tysseling et al. 2010). Briefly, a 27-gauge needle bent at 90° was inserted into the muscle belly, then the electrode wires were threaded into the needle and the needle was withdrawn, the two distal end of the electrode wires were then tied in a knot to secure the electrode within the muscle and the superfluous wire was trimmed. The EMG electrodes were implanted approximately 2 mm deep and approximately 1 mm apart, electrode placement was verified by electrical stimulation through the implanted wires. A pair of reference electrodes was inserted underneath the skin, in the region above the gluteus maximus muscle. Electrodes placement was further verified by dissecting the implanted muscles after termination of the animals. The skin was closed with sutures (5 − 0, Vicryl, absorbable, ETHICON). At completion of the surgery, 4 ml lactated ringer solution was injected subcutaneously. Then the rats were transferred back to a recovery cage placed on a heating pad and monitored for recovery.

Intramuscular hypertonic saline injection to induce nociception

Literature (Bagues et al. 2014; Hoheisel et al. 2005; Ro et al. 2003; Taguchi et al. 2008) and a pilot study (unpublished data) demonstrated that injection of 100 µl hypertonic saline into the rat’s MF muscle caused mild-to-maximum moderate pain responses based on the Rat Grimace Scale (score ranged: 0.5-1) (Leung et al. 2016; Miller et al. 2016). Therefore, in this study nociception was induced by injecting 100 µl hypertonic saline (5.8%) solution randomly into either the left or right MF muscle between the L4 and L5 vertebral levels using an insulin syringe according to the method described by Taguchi et al. (Taguchi et al. 2008). In brief, while the rats were under ultra-short isoflurane anesthesia, the needle was advanced into the muscle beside the spinous process until it contacted the bone of the vertebral arches. Then the needle was withdrawn for 1 mm to release the saline solution into the muscle.

In vivo measurement and data analysis

Before and after hypertonic saline injection, in vivo measurements were performed to collect locomotion data and EMG signals, while the rats were trotting on the treadmill at a fixed speed of 0.5 m/s (Fig. 1b). Kinematics data and EMG signals were synchronized by an electronic trigger pulse to the controller (Digital Sonomicrometer, Sonometrics, London, ON, Canada).

Kinematics

Two-dimensional videos of treadmill locomotion were recorded using a high-speed camera (A602f, Basler, Ahrensburg, Germany) placed above the treadmill (Fig. 1b). Videos were sampled at 200 frames/s and recorded at a computer hard drive with custom software (Labview, National Instruments, Austin, TX). Skin markers were placed on the rat’s spine at the L2 and S1 spinous processes, as well as on the pelvis at the left and right iliac crests (Fig. 1c).

Electromyography

EMG signals of MF and ML were amplified (1250×, common-mode rejection ratio > 100 dB), filtered (10–1175 Hz) and sampled (3123 Hz). Band-pass digital filters (100–1000 Hz, 3rd order zero-lag Butterworth) were applied for signal processing to remove movement artifacts and treadmill noise. The EMG linear envelope was computed as the magnitude of the discrete-time analytic signal calculated by the Hilbert transform and low-pass filtered (25 Hz, 2nd -order zero-lag Butterworth).

Data analysis

Videos within 5 min (Paintal 1960) after hypertonic saline injection were analyzed in DeepLabCut (Nath et al. 2019) to obtain the time series of the segmental angle data between the tracked markers (lumbar angle: marker L2 to S1; pelvic angle: marker left iliac crest to right iliac crest). As shown in Fig. 1c, the lumbar and pelvic angles were measured with reference to a horizontal line with the positive direction to the right. Extreme values in the segmental angle data (pelvic angle data beyond the range of 250–300 degree, lumbar angle data beyond the ranges of 0–30 degree and 330–360 degree) were removed and interpolated using the Piecewise Cubic Hermite Interpolating Polynomial method, then the angle data were low-pass filtered (5 Hz, 3rd order zero-lag Butterworth). The pelvic angle was used to separate stride cycles, and the start of the stride cycle was defined as the video frame in which the pelvic angle was minimal. Stride cycles of trotting at constant speed (cycle duration range: 0.2–0.4 s) were used for data analysis, galloping gaits and strides with forward-backward acceleration or left-right swing on the belt were excluded from analysis. Subsequently, the angle data were normalized to the stride cycle duration and interpolated to 100 data points. EMG data were also time-normalized to 100 time samples per stride. For each rat, a mean across stride cycles from the same measurement session was calculated, and the amplitude of the EMG envelope for each normalized time-point was normalized to the maximum value of the mean EMG recorded during baseline.

For each rat, changes in lumbar and pelvic angle over the stride were calculated based on the time-normalized segmental angle data. The offset of the angle data possibly caused by asymmetry in marker placement was removed by subtracting the mean of the time series. The spine angle was defined as the relative angle between the lumbar and pelvic angles. The timing of the peak pelvic angle was described as the percentage of the stride cycle at maximum pelvic angle. Variability of angle changes was expressed as the mean of the standard deviations across stride cycles for each time point. Movement asymmetry was calculated as the standard deviation of the differences between corresponding points in the first and second half of the spine angle curve, divided by half of the peak-to-peak difference of the curve. The peak and minimum EMG amplitude were defined as the maximum and minimum value in the normalized EMG envelope, respectively. The average EMG activity was calculated as the mean of the normalized EMG envelope. The mean of the standard deviations across stride cycles was used to indicate the variability of EMG and was calculated as described above for the kinematics. Each variable was averaged within rats.

Statistical analysis

EMG data from specific channels were excluded when there was a bad signal-to-noise ratio, electrode malfunction or inappropriate placement of the electrodes. Paired t-tests were performed to assess the effects of nociception on kinematics (pelvic/lumbar/spine angle change and variability, movement asymmetry), MF and ML EMG activity (peak amplitude, minimum amplitude, mean amplitude, variability). All statistical analysis were performed using MATLAB R2021a (MathWorks, Inc., Natick, MA, United States). Results were considered significant when p < 0.05, Cohen’s D was calculated as effect size. Data are presented as mean (SD).

The adopted statistical analysis deviated from the preregistered analysis plan (two-way repeated measures ANOVA). Two interventions have been preregistered, but only one intervention (hypertonic saline injection to induce nociception) is reported here. The other intervention (intervertebral disc injury, which was performed after the hypertonic saline injection) and the interaction between these two interventions will be reported later.