Early pain in females is linked to late pathological features in murine experimental osteoarthritis

Animals

Adult male and female C57BL/6J mice (n = 10 per group and sex; Janvier Laboratories, Le Genest-Saint-Isle, France), 10–14 weeks-old, were used in this study. Animals were housed, five per cage, under standard environmental conditions with food and water provided ad libitum. Nesting material and two prefabricated refuges (Igloo) per cage were provided as environmental enrichment. Animal order was blindly and randomly assigned by an operator who was not involved in the experiments. During the outcome measurement, blinding was not possible due to obvious phenotypic differences between the sexes. Data analysis was performed blindly. All animal procedures were designed and conducted in accordance with a protocol approved by the Ethical Committee of Animal Research of Radboud University and by the Dutch Central Committee for Animal Experiments (DEC approval n° 2015-0014; AVD10300 2015 115). The experiments followed the guidelines for the Care and Use of Laboratory Animals of the European Union Directive (2010/63/EU) and Dutch regulations (EC2013-235). The experiment was performed in two parts to prevent subjecting the mice to more than two behavioral studies, and a total of 40 mice were used (20 per experiment). As this was an exploratory study, sample size was determined based on previously unpublished data. In addition, 30 mice were used for the CiOA gait experiments database (described below). In total, 70 mice were used in the current study.

CiOA Model

Experimental OA was induced as described previously (Van der Kraan et al., 1990). Briefly, mice received two consecutive intra-articular injections in the right knee joint with one unit of collagenase type VII (Sigma-Aldrich, St. Louis, MO, USA) on day 0 and day 2 (Fig. 1). Consecutive injections were performed by the same operator. These injections resulted in local instability of the knee joint, evolving to an OA-like phenotype characterized by cartilage destruction, ectopic bone formation, and chronic synovial activation. Mice were sacrificed by cervical dislocation under isoflurane anesthesia, eight weeks after the first collagenase injection. Control groups were not included, as naïve (Blaney Davidson et al., 2006) or saline-injected (Van der Kraan et al., 1990) animals do not naturally develop changes in joint structure. Throughout the experiment, both the research team and the veterinary staff monitored the welfare of the animals daily. Animal health was monitored by weight, food and water intake, and a general assessment of activity, panting, and fur condition. In all of the experiments the animals did not reach the human endpoint (for detailed humane endpoint criteria, see Supplemental methods).

Experimental design

Pain-related behavior was evaluated weekly using four different methods to compare multiple aspects of pain between males and females. To prevent a higher degree of agitation from handling, and therefore a higher variability in the outcome, two pain measuring methods per experiment were performed on the same animals. For cohort 1, mechanical allodynia with the von Frey filaments was assessed, followed by weight-bearing in the affected joint with the incapacitance tester (IC), as previously described (Blom et al., 2020). For cohort 2, gait analysis using the Catwalk XT (Noldus, The Netherlands) was performed first as the more passive measurement, followed by mechanical hyperalgesia in the affected joint with the pressure application measurement (PAM; Ugo Basile, Italy) (Fig. 1). Due to the Catwalk apparatus requirements, the nociceptive behavior measurements for the second experiment were performed in a dark room, with red light illumination. For all pain measurements, extreme care was taken to ensure the reliability of the tests. Animals were trained for a period of two weeks before model induction. For the first week, mice underwent a handling/training procedure to get accustomed to the experimental settings. For the second week, three baseline measurements were obtained on alternate days before the start of the experiment: days −7, −3, and day 0. Mice were placed in the procedure room with the person performing the measurements 30 mins prior to the start of the experiment. For standardization purposes, individuals performing the procedure, the starting time, and the procedure order were kept constant throughout the experiment. After model induction, pain-related behavior was measured on days 3, 7, and weekly after that, until day 56.

To compare the pain progression between males and females over time, baseline correction was performed per individual mouse. The mean baseline was calculated and subtracted from each measurement, which set the normalized baseline value at zero. To compare histological parameters and pain per individual mouse, the area under the curve (AUC) was calculated using the raw data and each baseline threshold (min and max values), taking into consideration the peak of nociceptive behavior and whether it was positive or negative. Early and late AUC were calculated in respect of the periods between days 0 to 21, and 28 to 56, respectively, using software Prism 6 (GraphPad software, San Diego, CA, USA). These periods were determined based on the baseline recovery time in which the majority of animals were back to baseline threshold after the initial pain peak for all the parameters evaluated.

Nociceptive behavior assessment

Statistical analyses

Results are presented as mean ± standard error of the mean (SEM). The normality of all data was assessed using the Shapiro–Wilk test. Nociceptive behavior was analyzed by repeated measures (RM) two-way ANOVA, followed by the Bonferroni multiple comparison post-test for comparison between sexes and to the baseline. The P value was corrected for multiple comparisons. Male versus female differences in histology were assessed using the Mann–Whitney test. All statistical analyses were performed using the Prism 6 software (GraphPad software, San Diego, CA, USA), except for the correlation tests which were analyzed by Pearson correlation analysis using SPSS (IBM Corp., Armonk, N.Y., USA). Significant differences were considered as p < 0.05.

Mechanical allodynia and static weight-bearing symmetry following CiOA (cohort 1)

In cohort 1, mechanical allodynia was assessed with the ipsilateral foot pad using the von Frey filaments; however, this sensation was not observed in male or female subjects (Fig. 2A). Significant results were observed in males on day 49 compared to baseline, as well as in males compared to females; however, these differences were not indicative of mechanical allodynia. Within the same cohort, weight bearing asymmetry was more obvious during the early phase of CiOA in both males and females when compared to the later phase. Two-way ANOVA analysis showed that sex had a significant effect (p = 0.0118) with females displaying lower weight bearing in the affected hind limb than males.

Both sexes avoided loading support in the ipsilateral leg on days 3 (M 6.4 ± 9.9; F 19.6 ± 15.1) and 7 (M 7.5 ± 8.9; F 16.3 ± 7.0), with only females still showing impairment on day 14 (M 1.1 ± 8.6; F 6.7 ± 8.0). When comparing the magnitude of response between males and females, females displayed a lower weight bearing capacity on day 3. Females avoided bearing 20% of their weight on the affected limb compared to 6% in males (p < 0.001, Fig. 2B). Despite the differences in weight bearing, histological analysis of the joint at endpoint showed no significant differences between males and females in cartilage damage (M 10.7 ± 8.0; F 8.3 ± 5.8, Fig. 2C), osteophyte formation (M 10.3 ± 6.0; F 12.3 ± 3.6, Fig. 2D), and synovial alterations (M 87.6 ± 51.4; F 56.0 ± 27.3, Fig. 2E and M 5.2 ± 2.2; F 7.7 ± 3.8, Fig. 2F). Three males had joint dislocations compared to one female. These results indicate that females may experience more pain during the early phases of the disease or have different compensatory behaviors to relieve the affected joints, despite a similar degree of pathology at the endpoint.

Mechanical hypersensitivity in CiOA mice (cohort 2)

In cohort 2, male and female mice exhibited increased sensitivity in the affected joint on day 7 after the induction of OA. Males and females reacted to a force of 200.6 ± 99.3 gf and 95.4 ± 89.0 gf less than before model induction, respectively (Fig. 3A). Females recovered to near baseline force values by day 14. However, in males, the affected joints remained in a hypersensitive state during the full course of CiOA. Two-way ANOVA analysis showed that sex (p = 0.0063), time (p = 0.0008), and the interaction between them (p = 0.0037) had a significant effect on pain levels. The differences between males and females were present on days 14 (p < 0.01), 21 (p < 0.05), and 35 (p < 0.01), suggesting that males experienced increased sensitivity in the affected joint when compared to females. As opposed to the previous experiment, histological analysis revealed that males presented more cartilage damage as measured by the OA score (Fig. 3B, M 18.8 ± 8.0; F 7.8 ± 6.1; p = 0.0051), as well as more joint dislocations (six males versus two females) on day 56. No differences were observed in the number of osteophytes (M 15.1 ± 5.6; F 12.2 ± 6.0; Fig. 3C). Males tended to have a thicker synovium (124.5 ± 57.4 versus 74.0 ± 39.3, Fig. 3D) than females; however, this difference was not significant ( p = 0.063). The synovial cellularity remained similar between the sexes (M 12.5 ± 3.4; F 13.7 ± 4.4; Fig. 3E). These findings demonstrate that the higher and more prolonged hypersensitivity of the affected joints in males coincided with more severe histological parameters.

Spontaneous pain during voluntary gait following CiOA (cohort 2)

Cartilage damage was shown to be more severe in males, as mentioned in the previous section (Fig. 3B—Cohort 2). Males used the affected paw less often during early CiOA compared to baseline (Fig. 4A, D3, p < 0.01; D7, p < 0.001), which translated into a smaller print area on days 3 and 7. Males displayed a smaller print area than females on day 7. In addition, males also showed decreased weight support compared to baseline (max contact max intensity; Fig. 4B, D3, and D7, p < 0.01), suggesting mechanical allodynia. This impairment recovered to baseline values by day 14. In contrast to the males, females did not show significant changes compared to baseline in these footprint parameters at the early time points. However, females did show compensatory behaviors in the contralateral paw print area (Fig. S1A) over multiple time points during early and late phases. Two-way ANOVA analysis showed that sex had a significant impact on the affected paw print area values, with males displaying a lower print area than females (p = 0.0424). For the max contact max intensity, no significant effect of sex was observed.

Spatiotemporal parameters indicated that males experienced more nociceptive behavior in the early phases as indicated by a significantly longer swing time of the affected paw compared to females on day 3 (Fig. 4C). Males also displayed earlier weight bearing compensation by the contralateral hind paw (TDS; Fig. 4D) on day 7 (p < 0.05). Stand (Fig. 4E), single stance (Fig. 4F), and IDS of the affected hind limb (Fig. 4G) did not show significant differences between males and females. However, stand, IDS, and TDS two-way ANOVA analysis demonstrated that sex alone did not have an influence, but when interacting with time, the results were significant (p = 0.0065, p = 0.0191, and p = 0.0035, respectively), indicating that pain development over time in CiOA is different between males and females. The duty cycle, which is the ratio of the stand phase of the whole step cycle was significantly smaller in the CiOA limb in both males and females compared to baseline on day 3 (Fig. 4H, p < 0.001). Males had a significantly decreased duty cycle on days 49 (p < 0.05) and 56 (p < 0.01); however, no significant sex differences were observed.

Most of the changes in CiOA were observed in the early stages of the disease; however, CiOA seems to have promoted comparable adaptations in gait strategies on day 56 in males and females. Both sexes displayed a lower stand time on the affected paw (Fig. 4E), as well as a decreased IDS.

Gait analysis from the present study indicated that differences between males and females may be parameter-dependent, indicating differences in nociceptive-related compensatory mechanisms.

Association between nociceptive behaviors and histological parameters

To investigate whether sex is a determinant in the link between histological parameters and pain, data were analysed on an individual mouse level, disaggregated by sex. The area under the curve (AUC) of each pain parameter per individual mouse was calculated and used in correlation analyses to examine the relationship between the OA parameter and pain behavior. The number of significant correlations was compared between males and females. The AUC was first calculated for the entire time course of CiOA, since disease pathology at the endpoint is a result of events that occurred throughout the duration of the disease. This correlation analysis yielded 17 positive correlations for females, with r values between 0.934 and 0.642 (Fig. 5, Pearson r with significant levels p < 0.05 and p < 0.01), distributed over seven out of the 10 pain parameters. Males displayed only three positive correlations (r values: 0.645, 0.676, and 0.748, with significant levels p < 0.05), within two pain parameters, max contact max intensity, and swing time.

Most of the pain-related behavior was observed in the early phase; therefore, the correlation analysis was divided by the AUC of the early (day 0–21) and late phases (day 28–56). Remarkably, in the early phase, females showed 15 positive associations, with r values between 0.635 and 0.882, distributed over eight of the 10 pain parameters. The pain was associated more closely with hard tissue changes (cartilage damage and dislocation, and osteophyte number), in contrast to one positive association with synovial thickness. Males showed no associations between pain and hard tissue changes and displayed only two positive correlations with the synovial thickness. In the late phase comparison, females also showed positive associations; however, these were fewer than in the early phase, with 10 significant correlations in five of the 10 pain parameters. Females also displayed one negative correlation. Males showed only two negative correlations, suggesting that the number of osteophytes and synovial cellularity in the late phase are inversely proportional to swing time alterations.

Positive correlations indicate that the increase in two variables is proportionally related; and, these results indicate that the level of pain-related behavior highly correlates to histological parameters in females, while in males, this correlation was nearly absent. The complete data of the calculated correlations and corresponding p-values per parameter are shown in the Table S1.

Retrospective correlation analysis of pre-existing CiOA data

We re-analyzed pre-existing data (Table 2) on CiOA gait analysis using the catwalk according to the above-described strategy to validate our findings. Cartilage damage assessed by histopathological scoring displayed a high degree of variation between sexes and experiments. For the experiment including males and females (Dataset 1), males showed slightly more severe cartilage damage compared to females, however, the difference was not significant (p = 0.0524, Fig. 6A). Males also presented more knee joint dislocations than females (six in males and three in females). In the experiment with only females (Dataset 2), cartilage damage and the number of animals with dislocation (seven out of 10 animals) were similar to the males in experiment 2 (Fig. 3). When combining the data from the pre-existing dataset and the experiments (females: n = 30 and males: n = 20), the correlation analysis from the AUC of the pain parameters and cartilage damage and dislocations resulted in significant positive correlations in females only. Females displayed four moderately positive correlations with r values between 0.376 and 0.559 (Fig. 6B, Pearson r with significant levels p < 0.05 and p < 0.01) for females, distributed over two of the six evaluated pain parameters. In evaluating the effects of the early and late phases of the disease, females showed four positive correlations, with higher r values, between 0.500 and 0.610 in the early phase (day 0–21) (Fig. 6B, Pearson r with significant level p < 0.01). The late phase correlations showed no associations between pain and damage in either males or females, with males showing no significant correlations in any of the periods or parameters investigated.

The inclusion of data from two previous studies in CiOA, which were subjected to the same association analysis, substantiated our primary findings that the association between the histological parameters of OA is present in females and is nearly absent in males. The complete data of the calculated correlations and corresponding p-values per parameter are shown in Table S2.