Adult male C57BL/6J mice [total n = 152 mice, 10–12 weeks, weight 26 ± (SD) 2.2 g] were kept in a 12/12 h day/night cycle with ad libitum access to food and water under standardized specific-pathogen-free (SPF) conditions (Figure 1A, General). According to their experimental group, mice were housed together (five per cage) to minimize intergroup effects. The experimental group allocation was random, and a blinded analysis of video-based behavior assessments was performed. Blinding to the withdrawal reflex-based behavioral assays and the pain model was not possible due to the testing conditions and visible signs of both pain models, for example, inflammatory reaction or sutures. Finally, animals were euthanized by an overdose of carbon dioxide at the end of the cohort-specific observation period.

All procedures were performed on the right hindpaw of mice. Mice were initially anesthetized with 5% isoflurane in 100% oxygen; anesthesia was maintained with 1.5–2.0% isoflurane delivered through a nose cone during the whole procedure. No analgesics were administered.

Paw incision (Figure 1A, Pain Model) was based on the procedure introduced by Brennan et al. in rats and adapted to mice as previously described (14, 20). In brief, the entire right hindpaw was disinfected with 100% ethanol and Betadine® (povidone-iodine). No antibiotics were administered. On the plantar aspect, epidermis, dermis, fascia, and musculus digitorum brevis were incised longitudinally with a scalpel (No. 11, 0.5 cm). Additionally, the muscle was elevated and short-term retracted with forceps (Dumount #5) (21). A mattress suture of 7-0 Prolene within the skin was used to close the incision. The wound was covered by Betadine® using a cotton swab. The sutures were removed at the end of 2nd day under light isoflurane anesthesia. Sham-treated mice (anesthesia only, the same time duration) were used to control the surgical incision (Sham INC).

As shown in previous studies, a CFA reaction was initiated by injecting 20 μl complete Freund's adjuvant (CFA, Sigma-Aldrich GmbH, Germany) subcutaneously (s.c.) into the plantar aspect of the right hindpaw with a 27-gauge Hamilton microsyringe (15) (Figure 1A, Pain Model). Sham-treated mice received an s.c. application of sterile 0.9% saline (20 μl) in the same aspect of the paw (Sham CFA).

Ly6G⁺ NGs were systemically depleted through an intraperitoneal (i.p.) injection of a purified anti-mouse Ly6G/GR-1 antibody (7.5 μg, RB6-8C5 clone, eBioscience, San Diego, CA, USA). Starting 2 days before the induction of pain models, the Ly6G/GR-1 or IgG2b (as vehicle control) antibodies were applied every other day throughout the entire experimental period (16) (Figure 1A, Injection Scheme) in both pain models. Untreated mice (no injection) were used as a separate control.

Inflammation is characterized by five cardinal signs: calor, tumor, rubor, dolor, and function laesa.

The results of the macroscopic parameters and MPO expression were analyzed by Kruskal–Wallis and Dunnett's multiple comparison tests. To achieve a better comparability of the individual behavioral data across the different pain modalities for this longitudinal study, mean changes in percent were calculated for NEP, PWT, and PWL (raw data are reported in Supplementary Figure 1). The ratiometric analysis of MEP expression is necessary as the weight of animals increases during the experimental period, and a comparison of the individual parameters on raw data is only possible to a limited extent. The Kruskal–Wallis and Mann–Whitney rank-sum tests were used for behavioral experiments followed by post-hoc tests (a two-tailed Dunnett's test). After the Friedman and Kruskal–Wallis test, multiple comparisons were performed by a Dunnett's test (to the sham group) for non-parametric raw data. Spearman coefficients (two-tailed) were used to determine relationships between the multiple parameters. Correlation analyses were performed after first computing the mean of side-by-side replicates and then analyzing those means. The value of p < 0.05 was accepted as significant. All results are expressed as mean ± SEM.

To assess the Ly6G⁺ neutrophil accumulation and the efficacy of global depletion via Ly6G/GR-1 antibody, the expression of the NG-specific enzyme MPO was determined in both pain models in skin and muscle samples of ipsilateral hindpaws (cohort 1). In the IgG2b-treated (vehicle) group, MPO peaked significantly at 6 h and 1 day in CFA but only 1 day post-INC the skin (Figure 2A) and muscle (Figure 2B). Thereafter, progressive recovery of NG accumulation was observed in both pain models, which reached the baseline values 2 days post-injury. Ly6G/GR-1 antibody treatment significantly reduced MPO in both tissues and pain models (Figures 2A,B).

Edema formation and skin temperature of the ipsilateral hindpaw were determined over two weeks in cohort 2. In the acute period, ipsilateral hindpaw thickness increased significantly in both pain models under IgG2b control treatment with slightly different durations (until 1 day after INC and until 2 days after CFA injection) (Figure 2C). In IgG2b-treated mice, skin temperature of ipsilateral hindpaws increased significantly in INC mice in the acute phase but only until 6 h post-incision (Figure 2D). In contrast, skin temperature significantly increased for at least 6 days in the CFA group treated with the control antibody (Figure 2D). NG depletion using Ly6G/GR-1 antibodies significantly decreased the incision-induced thickness and temperature of ipsilateral hindpaws 6 h following the surgery relative to their vehicle controls. However, both parameters were not affected in NG-depleted CFA-treated animals (Figures 2C,D).

Classical reflex-based withdrawal assays were used to determine the thresholds to mechanical stimuli (PWT) and the withdrawal latencies to heat stimuli (PWL) at the hind paw. In IgG2b-treated mice (vehicle group), PWT and PWL were significantly reduced by days 2 and 3 in INC animals, respectively (Figures 3A,B). PWT and PWL were significantly decreased by day 6 after CFA injection in IgG2b-treated mice (Figures 3C,D). INC and CFA sham-treated mice showed no changes in PWT and PWL (Figures 3A–D and Supplementary Figure 1). NG depletion slightly worsened PWL on day 1 after INC (^†††p < 0.001), whereas PWT was unchanged in comparison to those injected with IgG2b. After CFA injection, the PWT in NG-depleted mice was only significantly decreased until day 2 in comparison to sham-treated mice and was significantly ameliorated at day 6 in comparison to IgG2b-control treatment. Heat hypersensitivity was ameliorated from day 3 after CFA injection compared to IgG2b-treated mice.

To move beyond classically evoked behavior assays, we employed a novel video-based assessment of non-evoked pain behavior (NEP) at rest. In this assay, NEP is represented by a decrease in the print area at rest of the ipsilateral paw. NEP was significantly reduced in both pain models starting in the acute phase (4–6 h) and remained significant until day 6 in the IgG2b-control mice group (Figures 3E,F). Ly6G⁺ neutrophil depletion did not modulate NEP in both pain models. Sham-treated mice showed no changes in NEP (Figures 3E,F and Supplementary Figure 1).

In addition to NEP, movement-evoked pain behavior (MEP) also represents a relevant symptom in different pain diseases. With the CatWalk XT system, static and dynamic parameters of voluntary ambulatory gait are evaluated under standardized conditions to assess MEP. Two static (print area and stand duration) and two dynamic (swing speed and stride length) gait parameters known to be altered in unilateral pain models were selected to assess MEP (Figure 4). Both static parameters and the dynamic parameter swing speed were altered significantly in the acute phase (4–6 h) post-incision in IgG2b-treated mice. However, in the further observation period, these profiles became diversified; the print area was decreased until day 6 (Figure 4A), stand duration returned to the baseline on day 1 (Figure 4B), and swing speed on day 3 after incision (Figure 4C). In addition, the stride length (Figure 4D) was not changed at any time point post-incision. In contrast to INC group, MEP was not detected in the acute phase after CFA (Figures 4E–H). The print area was significantly reduced at day 2 and 3, and stand duration only at day 3 (Figure 4E). Swing speed was significantly decreased from 1 to 3 days (Figure 4G). As in the INC group, the stride length was not modulated over time. For all pain-related gait parameters, baseline values were reached at day 14 at the latest in both models, except the print area in CFA mice. Ly6G⁺ neutrophil depletion did not significantly affect MEP in both pain models (Figure 4E).

We performed a correlation analysis of behavioral outcome parameters to investigate the possible coherence between the multidimensional pain modalities that could inform about shared mechanistic pathways. Because Ly6G⁺ neutrophil depletion only showed marginal effects on pain-related behaviors in both pain models, we limited our analysis to mice receiving IgG2b. First, we performed a correlation analysis between macroscopic parameters and reflex-based withdrawal assessments to investigate a link between them (Figure 5). A significant time-dependent negative correlation was found across all time points only for PWT and hindpaw thickness for INC, indicating a decrease in PWT associated with an increase in edema formation (Figure 5A). Hindpaw skin temperature did not correlate with any behavioral tests in INC although we noted a trend (p = 0.07) for a negative correlation with PWT (Figure 5B). While there was only a trend (p = 0.06) between PWT and paw edema in CFA-treated mice (Figure 5C), a significant correlation between skin temperature and PWT was observed (Figure 5D). Further, NEP was negatively correlated with skin temperature in the CFA group (Figure 5D). Interestingly, responses to heat stimuli did not correlate with any inflammatory parameter. The correlation analysis between the reflex-based behavioral tests and the NEP showed significant positive correlations across all modalities and both pain models (Figure 6). When we correlated PWT and NEP with MEP parameters, a different picture emerged (Figure 7). Here, we found significant positive correlations of PWT and NEP with the dynamic parameter “swing speed” in the INC model. Other gait parameters did not show significant correlations.

The assessment of “pain” in rodents is a multidimensional approach and presently a matter of ongoing debate. Despite significant preclinical pain research efforts, a translational gap exists between pain-related outcomes in rodents and humans (1). This leads to an insufficient understanding of mechanisms of pain entities and the unsuccessful development of novel therapeutic options. To bridge the translational gap between preclinical and clinical pain measures (1, 6, 10, 27, 28), we used here novel video-based paradigms based on voluntary behaviors in mice to assess the behaviors that more likely resemble pain-related outcomes in patients. “Non-evoked pain” is a concept based on the modulations of various natural and rodent-specific behaviors like grooming, burrowing, weight-bearing, ultrasound vocalization, facial expression, emotional, and cognitive aspects, to name a few (10). It is also reflected by the scoring of the guarding behavior (21, 29, 30) and the spontaneous foot lifting (SFL) of the affected paw in unilateral pain models (31, 32). Problematic ones might be a scoring system prone to be biased by subjective rating. Thus, we developed a novel paradigm to detect guarding behavior as voluntary and more objectively by investigating the reduced print area of the injured hindpaw at rest in mice (13). Guarding is defined as “behavior […] aimed at preventing or alleviating pain” (33); as such, it could be a behavior to avoid mechanical stimulation as a result of contact with the ground or be a sign of ongoing pain or may be a mixture of both as described previously (33). Our results here demonstrate a significant guarding behavior based on this approach that peaked early after onset in both pain entities, lasted until day 6, and showed complete remission on day 14. The acute phase peak of guarding behavior was also described in the studies performed with subjective scoring in mice (21, 29) and rats (34). However, in contrast to rats, the determination of guarding behavior with the “scoring” assessment in mice is more difficult due to anatomical differences, higher mobility of mice, challenges in acclimatization to the experimental setting, and the impact of an observer (21). Therefore, only a few studies exist in the literature to describing non-evoked behaviors in mice in the incision (4, 21, 35) and the CFA model (36, 37), mainly due to methodological problems of this behavior assay. Using a video-based, standardized assay that can detect even small footprint-based area changes and underlies objective measures and no scoring, many disadvantages of the “scoring” method are eliminated, especially the experimenter bias. However, we have to admit that the time course of our footprint-based approach has a similar time course to the reflex-based pain behaviors; this suggests that the footprint approach might be at least a combination of non-evoked behavior as well as avoiding touch/pressure to the underlying ground. A pure “non-evoked” behavior that lacks the flaws of the scoring system after incision needs further investigation. Nevertheless, such an avoidance of behavior might as well be relevant for better translation to the clinics.

In line with previous findings in rats after plantar incision (38) and CFA-treated rodents (26), we were able to detect an antalgic gait, which is characterized by a decreased swing speed associated with a reduction of the stand duration and the print area of the hindpaw for both pain models early after onset injury. This type of movement-evoked behavior in mice might be of major importance to be studied in more detail. Recently, it was mutually agreed with patients and many stakeholders involved in perioperative pain management, pain-related physical function (like moving out of bed, performing physiotherapy, etc.) is the most important domain to be assessed in clinical acute perioperative pain (39). Future studies need to identify relevant mechanisms of movement-evoked pain in mice to develop effective treatment options.

A correlation analysis of pain modalities may inform about the entity- and modality-specific mechanisms. We show here a significant correlation between mechanical and heat hypersensitivity in both pain entities. According to previous studies, it is clear that mechanical and heat hypersensitivity underlie different pathophysiological mechanisms (40), these processes seem to correspond time-wise. Furthermore, both modalities were correlated with non-evoked pain behavior although the correlation was most pronounced with heat hypersensitivity. Further, movement-evoked behaviors and withdrawal reflex-based assessments do not correlate, suggesting that the underlying pathophysiological processes differ. These differences are already evident in the stimulus type; however, they highly suggest that movement-evoked mechanisms have been missed to be investigated in the past in preclinical studies. While reflexive paw withdrawal is evoked by punctual stimulation of distinct nociceptors/sensory types, the ambulatory activity of the hindlimb during movement reflects peripheral inputs that are the result of the integration of multiple tissues and nociceptor/sensory types throughout the paw and hindlimb. In addition, such a voluntary behavior might involve supraspinal responses and therefore was better related (than reflex-based approaches) to a more complex construct of pain in humans. Together, this might explain why translating results obtained with “old-school” behavioral assessments do not translate well to the clinic.

Currently, conflicting reports exist on whether pain-related behaviors are affected by antibody-mediated peripheral depletion of Ly6G⁺ neutrophils in C57BL/6 mice. While Carreira et al. demonstrated a reduction of mechanical hypersensitivity after Ly6G⁺ neutrophil depletion post-incision (17), others reported no change in mechanical thresholds accompanied by a slight worsening of heat hypersensitivity (16) or no effect on neither mechanical nor heat hypersensitivity after incision (15). In addition to these conflicting results, all findings have been based on reflex-based approaches only, reducing the generalizability and significance of the results.

Here, we show that Ly6G⁺ neutrophil depletion does not change mechanical hypersensitivity after incision but might have a transient, mild effect on incision-induced heat hypersensitivity. This is in contrast to the study of Carreira et al. who identified a major role of Ly6G⁺ neutrophils for mechanical hypersensitivity after incision (17). In addition, the results in the incision model differed from a previous study using the CFA model (15) where, in our hands, mechanical hypersensitivity was unchanged, but the recovery of heat hyperalgesia was accelerated. These conflicting results between the studies within the same pain model might be attributed to different timing and routes (i.v. vs. i.p.) of Ly6G/GR-1-antibody application as well as the differences in pain model induction characteristics and, as described above, bias in assessment. It is known that experimental parameters like incision depths are particularly crucial for the magnitude of immune response and relevant for nociceptor activation in dermis vs. muscle; thus, pain-model induction influences the outcome (41, 42). It is also noteworthy that edema and hypersensitivity do not necessarily correlate with incisional (16, 29) and inflammatory pain (29, 43). The different roles of Ly6G⁺ neutrophils for pain behavior in different pain models might be explained by a predominant secretion of either pro- or anti-inflammatory mediators in each pain model (16). This dichotomous role (anti- vs. pro-inflammatory) has been reported previously in different pain models (15, 44).

Antibody-based depletion of immune cells is a standard method to investigate their disease-specific role but is largely determined by antibody selectivity. To achieve maximum comparability with the previous studies investigating the role of Ly6G⁺ neutrophils in the context of pain in rodents, the most widely applied antibody (Ly6G/GR-1, clone RB6-8C5) was used in this study. However, this antibody may deplete, albeit marginally (15), various subtypes of monocytes and wound macrophages that have an impact on inflammatory processes and therefore must be considered when interpreting the data shown here (45).

Here, we show that Ly6G⁺ neutrophil recruitment plays a minor role in developing and maintaining hypersensitivity to mechanical and heat stimuli and is also done without any contribution to non-evoked and movement-evoked pain-related behavior associated with incisional and inflammatory pain. These findings are partially consistent with some reports highlighting the role of different immune cell types in both pain models that modulate sensitization nociceptors by releasing inflammatory mediators. However, as all previous studies were exclusively based on reflex-based behaviors, these results might not be of major relevance to be translated to the clinic. Indeed, due to the assessment of non-evoked and movement-evoked pain-related behavior with our multidimensional behavior approach, we clearly show that the migration of Ly6G⁺ neutrophils does not contribute to those pain-related behaviors, which translate to more clinically relevant outcomes in patients (e.g., physical function). Based on these data, peripheral Ly6G⁺ neutrophils in the skin and muscle tissue are not a promising cellular target for developing novel analgetic options in these two pain entities. In the future, a multidimensional behavioral approach lays the foundation for examining new and revisiting known mechanisms for their translational potential to gain novel perspectives, providing potential valuable information for translation to the clinic.