Intra-articular injection of collagenase in the knee of rats as an alternative model to study nociception associated with osteoarthritis

Experimental procedures were performed in accordance with the ethical guidelines for the study of experimental pain in conscious animals [34], and the European Communities Council Directive 86/609/EEC amended by the Directive 2003/65/CE and were approved by the Ethical Committee for health of Hospital de São João, Porto, Portugal.

Under brief isoflurane anaesthesia (5% isoflurane for anaesthesia induction, 2% for maintenance), an intra-articular injection was performed with the use of a Hamilton syringe, with a 26 G needle inserted through the patellar ligament into the joint space of the left knee. Animals received two injections, one on day 0 and another on day 3, with 25 μl of either sterile saline (control group), 250 U or 500 U of type II collagenase from Clostridium histolyticum (Sigma-Aldrich, St.Louis, MO, USA) dissolved in saline and filtered through a 0.22 μm membrane [20]. Animals were randomly assigned to each group before the first injection. Doses of collagenase were chosen based on a previous study that assessed nociceptive alterations in rats [35] injected with 500 U of collagenase. We tested the same dose and a smaller one, to evaluate if similar changes could be obtained with a smaller amount of exogenous collagenase.

Animals injected with saline, 250 U or 500 U collagenase (72 animals; n = 24/treatment) were randomly divided in groups that were sacrificed at different time-points: one, two, four and six weeks after the first injection (n = 6/group/treatment). Movement-induced nociception was evaluated by the Knee-Bend and CatWalk tests [13] on Day 0 (before the first injection), and at one, two, three, four, five and six weeks after the first injection, until each group’s endpoint. Testing was performed blindly, always by the same experimenter. The Knee-Bend test consists in counting the squeaks and/or struggle reactions in response to five alternate flexions and extensions of the knee joint, performed within the physiological limits of knee flexion/extension. The score of the test was determined by the type of reaction to each movement of the joint as follows: 0 - no responses to any kind of movement of the joint; 0.5 - struggle to maximal flexion/extension; 1 - struggle to moderate flexion/extension or vocalizations to maximal flexion/extension; 2 - vocalizations to moderate flexion/extensions. A maximal extension corresponds to placing the knee joint in an 180° angle; a moderate extension corresponds to an angle between 120° and 150°, approximately; a moderate flexion corresponds to an angle between 45° and 75°, approximately; a maximal flexion corresponds to totally bending the knee joint (corresponding approximately to an angle of 30°). The sum of the animal’s reactions, giving maximal values of 20, represents the Knee-Bend score, an indication of the animal’s movement-induced nociception. The contralateral knee was always tested first, in order to avoid an increase in the contralateral score arising from the manipulation of the injected knee. Results for both ipsilateral and contralateral knees were presented.

For the CatWalk test, animals were placed in a glass platform illuminated such as to reflect light only at the points of contact of the paw with the surface, resulting in a bright image of the paw print. Videos were acquired by a camera placed under the platform. The signal intensity depended on the paw area in contact with the platform and increased with the pressure applied by the paw. Random frames of the videos were analysed: three pairs of frames (one for each hind paw) with the animal walking and three frames with the animal standing still. For each hind paw, the number and intensity of pixels above a defined threshold were quantified (Image J 1.37, [36]) to determine the total paw print intensity (mean pixel intensity × number of pixels), allowing the comparison of the area/pressure applied by each paw. Results were expressed as the percentage of the total ipsilateral paw print intensity (%TIPPI) in the total intensity of both paw prints.

The CatWalk test was always performed prior to the Knee-Bend test, to minimize the effect of manipulation of the affected knee joint on the animals’ gait.

Knee diameters were measured to infer joint swelling as an indicator of inflammation resulting from i.a. injection of collagenase or saline. The diameter of both knees was measured with a manual calliper. Results were presented as the difference in knee diameter (ipsilateral-contralateral).

At one, two, four and six weeks after the first injection of saline, 250 U or 500 U of collagenase, animals were perfused with 4% paraformaldehyde (n = 6/group/treatment). Their ipsilateral DRG from lumbar segments L3, L4 and L5 were dissected, post-fixed in the same fixative (four hours) and kept in 30% sucrose with 0.01% sodium azide. DRG were serially sliced in 12 μm sections using a cryostat, with every 10^th section collected in the same slide. DRG were always cut longitudinally yielding 8 to 10 sections from each, on average. The injected knees were also dissected, post-fixed for 72 hours and then decalcified for 8 hours in a buffer containing 7% AlCl₃, 5% HCOOH and 8.5% HCl. The joints were then washed in 0.1 M phosphate buffer pH 7.2, and kept in 30% sucrose with 0.01% sodium azide until they were cut. Joints were cut into two approximately equal halves along the medial collateral ligament in the frontal plane. Three 10 μm frozen sections were cut from each half at 200 μm steps using a cryostat.

Knee joint sections were stained either with Haematoxylin and Eosin or by the Fast Green and Safranin-O method to evaluate the extent of the histopathological lesions. Slides were mounted with Eukitt (Kindler GmbH & Co., Baden-Württemberg, Germany) and images acquired with an Axioskop-40 microscope equipped with an AxioCam-MRc5 camera (Carl Zeiss MicroImaging GmbH, Jena, Germany).

Histological scoring was performed for the medial tibial plateau (MTP) of the three most severely affected sections based on the Osteoarthritis Research Society International (OARSI) recommendations for histological assessment of osteoarthritis in the rat [37]. Using an image analysis software (Image Pro-plus 5.0, Media Cybernetics, Rockville, MD, USA), the following parameters were evaluated, as fully described by Gerwin and colleagues [37] and briefly described here:

Slides containing every 10^th section of ipsilateral L3, L4 or L5 DRG of animals sacrificed one or six weeks after collagenase or saline injection were used for immunofluorescence reactions against TRPV1. DRG sections were rinsed in 0.1 M phosphate-buffered saline (PBS; pH 7.4), followed by PBS + 0.3% triton X (PBST), and incubated in PBST with 10% normal serum (1 h 30). Sections were then incubated with guinea-pig anti-TRPV1 diluted in PBST + 2% normal serum (1:250, Chemicon, Temecula, CA, USA), overnight at room temperature. After thorough PBST washing, sections were incubated with Alexa Fluor 568 donkey anti-guinea-pig secondary antibody diluted in PBST + 2% normal serum (1:2,000, one hour, room temperature, Molecular Probes, Eugene, OR, USA). Slides were then rinsed in PBST followed by PBS, mounted with Prolong Gold Antifade mounting medium (Molecular Probes) and cover-slipped. Negative controls were performed by following the same procedure without the primary antibody. DRG sections were viewed using a Zeiss Imager.Z1 fluorescence microscope (Carl Zeiss MicroImaging GmbH, Jena, Germany) and images were acquired using an AxioCam MRm with AxioVision 4.6 software (Carl Zeiss MicroImaging GmbH, Jena, Germany) at a 100x magnification. All neurons and all labelled cells were counted in one slide containing every 10^th section of each DRG. Data were presented as the percentage of neurons expressing TRPV1 in L3, L4 and L5 DRG.

A pharmacological evaluation was performed in different groups of animals at one, two, four and six weeks after 500 U collagenase injection using the opiate morphine, the local anaesthetic lidocaine and the NSAID diclofenac. Baseline values of nociception were determined before drug administration (t = 0 minutes). To confirm that the behavioural responses of OA animals in the CatWalk and Knee-Bend tests were induced by nociception, we evaluated if they could be reversed by a single administration of a non-sedative dose of morphine [7]. Morphine (6 mg/mL in saline, Labesfal, Lisboa, Portugal) was subcutaneously (s.c.) injected in the upper dorsum in a dose of 6 mg/Kg [7, 38] (n = 5), and drug effects on both tests were assessed 30, 60, 90, 120 and 180 minutes after administration. Control animals were also injected with 500 U of collagenase and received a s.c. injection of saline (n = 4). Behavioural assessment was performed after 30, 60 and 120 minutes. To determine if nociception in this model originates from knee articular tissues, we intra-articularly injected the local anaesthetic lidocaine in the knee of rats injected with 500 U of collagenase. Lidocaine (100 mg/mL in saline, Sigma) was administered in a dose of 5 mg, in a volume of 50 μL [7, 38] (n = 5). Drug effects on both tests were assessed after 10, 20 and 30 minutes. Control animals received an intra-articular injection of saline (n = 4). To assess if inflammation occurs during the onset and/or development of the model, we evaluated the effect of the NSAID diclofenac. Diclofenac (15 mg/mL in bidistilled water, Sigma) was administered per os (p.o.) in a dose of 30 mg/Kg [7, 38] (n = 5), and drug effects on both tests were assessed 30, 60, 90 and 120 minutes after drug administration. Control animals were also injected with 500 U of collagenase and received a p.o. administration of bidistilled water (n = 4). Data are presented as the percentage of changes in relation to baseline values for the time-point of peak effect for each drug and nociceptive test. Animals were randomly assigned to each group and testing was performed blindly. Post-administration time-points for behavioural assessment were determined in preliminary experiments.

Animals were randomly assigned to each group. Results are presented as mean ± SEM. The normality of all data was assessed by the Kolmogorov-Smirnov test. Behavioural data for movement induced-nociception in control, 250 U and 500 U collagenase-injected groups were analysed by two-way ANOVA, for factors time and group, followed by Bonferroni post-hoc test for multiple comparisons between groups; ipsilateral vs. contralateral comparisons in the Knee-Bend test were performed likewise. The pharmacological data were analysed by repeated measures ANOVA followed by Dunnett’s post-hoc test. For the histological assessment, measured parameters were analysed by one-way ANOVA followed by Tukey post-hoc test; scored parameters were analysed by the Kruskal-Wallis test with Dunn’s post-hoc test. Pearson’s correlation analysis was performed for data from control and 500 U collagenase-injected animals. The expression of TRPV1 was evaluated by the Mann–Whitney test. A P-value <0.05 was accepted as statistically significant.

Signs of inflammation were detected after collagenase injection, denoted by a dose dependent increase in the difference between the ipsilateral and contralateral knee diameter at one week, significant only for the 500 U dose (P <0.001, Figure 1), with ipsilateral joint diameters returning to contralateral values thereafter.

The Knee-Bend (Figure 2A-C) and CatWalk (Figure 2D-E) tests were used to evaluate loading- and movement-induced nociception following the i.a. injection of different doses of collagenase. In both tests and at all time-points, saline-injected control animals showed behavioural responses similar to those observed prior to the injection (day 0; 3.3 ± 0.3 for the ipsilateral Knee-Bend score; 51.0 ± 0.7% for the CatWalk). In collagenase-injected animals, behavioural changes were time-dependent, with the highest increase in nociception being observed one week after the first collagenase injection, and remaining higher than on day 0 or in control animals throughout the whole period of analysis for both collagenase doses (Figure 2). For the 250 U, ipsilateral Knee-Bend scores (Figure 2B) increased from 3.3 ± 0.4 on day 0 to 13.1 ± 0.9 on week 1 (P <0.001), slightly decreasing throughout time, but remaining significantly different from the contralateral knee and control animals until the sixth week (10.2 ± 1.7, P <0.001). For the 500 U dose, ipsilateral Knee-Bend scores (Figure 2C) increased from 3.2 ± 0.6 on day 0 to 15.7 ± 0.7 on week 1 (P <0.001), slightly decreasing throughout time, but remaining significantly different from the contralateral knee and control animals until the sixth week (12.3 ± 1.0, P <0.001). The increase in the ipsilateral Knee-Bend score for the 500 U dose at week 1 was significantly higher than the increase observed with the 250 U dose (P <0.05). The values obtained with the 500 U remained higher than those with the 250 U throughout time, although differences were not statistically significant. Likewise, in the CatWalk test, the %TIPPI showed a similar pattern for both doses, being always below the values observed on day 0 and in control animals, although not always reaching statistical significance (Figure 2D). Thus, for the 250 U dose the %TIPPI decreased from 49.3 ± 1.0 on day 0 to 41.9 ± 1.4 at week 1 (P <0.001), when the lowest values were observed, slightly increasing until the sixth week to a value of 44.4 ± 1.9, not significantly different from control animals. For the 500 U dose (Figure 2E) the %TIPPI decreased from 49.4 ± 0.8 on day 0 to 35.6 ± 2.4 at week 1 (P <0.001), when the lowest values were observed, and increased from then until week 4 (46.6 ± 1.9, n.s.), decreasing again until the sixth week (42.0 ± 1.5, P <0.05). As for the Knee-Bend test, the decrease in the %TIPPI for the 500 U dose at week 1 was significantly higher than for the 250 U dose (P <0.01), with 500 U values remaining lower at late time-points but not significantly different.

The effect of an acute administration of morphine, lidocaine and diclofenac on loading- and movement-induced nociception was evaluated at one, two, four and six weeks after injection of 500 U of collagenase (Figure 3), since this was the dose that induced nociceptive changes that better correlated with relevant histopathological alterations. All animals showed increased nociceptive behaviour before drug or vehicle administration (t = 0 minutes, baseline). The effect of a s.c. injection of saline, a p.o. administration of bidistilled water or an intra-articular administration of saline was evaluated in animals injected with 500 U collagenase to monitor the vehicle’s effects. We obtained similar results to those already published by our group for the monoiodoacetate (MIA) model of OA [38], with no changes being observed (data not shown).

Morphine was highly effective at all time-points of disease progression in the Knee-Bend test, fully reversing the nociceptive behaviour (Figure 3A, B). The maximal effect was observed between 30 and 90 minutes after injection, starting to revert thereafter. Thirty minutes after administration, morphine significantly reduced the ipsilateral Knee-Bend score (Figure 3A) by 95 ± 5% at week 1 (P <0.001), 93 ± 3% at week 2 (P <0.001), 95 ± 4% at week 4 (P <0.001) and 97 ± 3% at week 6 (P <0.001). In the CatWalk test (Figure 3B), morphine was effective one, two and six weeks after collagenase injection and maximal effect was observed from 60 to 90 minutes. Sixty minutes after administration, the %TIPPI increased by 40 ± 15% at week 1 (P <0.01), 30 ± 10% at week 2 (P <0.05), 7 ± 5% at week 4 (n.s.) and 42 ± 5% at week 6 (P <0.01).

Lidocaine was also effective in diminishing the nociceptive behaviour (Figure 3C, D) at all time-points. The maximal effect was observed 10 minutes after injection, starting to revert thereafter. In the Knee-Bend test (Figure 3C), lidocaine significantly reduced the ipsilateral Knee-Bend score by 50 ± 8% at week 1 (P <0.001), 53 ± 7% at week 2 (P <0.001,), 52 ± 9% at week 4 (P <0.05) and 62 ± 13% at week 6 (P <0.01). In the CatWalk test (Figure 3D), lidocaine increased the %TIPPI by 18 ± 4% at week 1, although this was not statistically significant, 40 ± 12% at week 2 (P <0.01), 31 ± 11% at week 4 (P <0.01) and 41 ± 11% at week 6 (P <0.01).

Diclofenac, on the other hand, was only significantly effective at week 1 (Figure 3E, F). Its maximal effect occurred 30 minutes after injection with a decrease of 21 ± 3% in the Knee-Bend score (P <0.05, Figure 3E) and an increase of 22 ± 5 in the %TIPPI in the CatWalk test (P <0.05, Figure 3F).

No histopathological damage in the knee joint could be found in saline-injected control animals at any time-point studied (Figure 4A). In contrast, dose- and time-dependent alterations were observed in collagenase-injected rats, which were always more pronounced in the MTP (authors’ observation, not shown). At week 1, a slight superficial loss of proteoglycan staining was observed in animals injected with both doses of collagenase (Figure 4B, F), as well as a high degree of a dose-dependent synovial inflammation (Figure 5B, C). Two weeks after collagenase injection, there was a dose-dependent thinning of the articular cartilage (Figure 4C, G), accompanied by focal chondrocyte disorganization, for the 250 U dose (Figure 4C), and by chondrocyte clustering and hypertrophy for the 500 U dose (Figure 4G). Synovial inflammation, although less pronounced than at week 1, was also observed, particularly with the 500 U dose. At week 4, some erosion of the articular surface was observable with the 250 U dose (Figure 4D). With the 500 U dose further signs of OA development were observed, such as the occurrence of small fissures and superficial erosion of the articular cartilage, along with chondrocyte clustering and hypertrophy (Figure 4H). Synovial inflammation was reduced for both doses and some small osteophytes could be observed. After six weeks, the 250 U dose induced loss of proteoglycan staining, focal chondrocyte disorganization, irregularity of the articular surface, and areas of cartilage matrix loss (Figure 4E). With the 500 U dose, clear signs of OA development were observed: extensive damage of the articular cartilage (Figures 4I and 6); areas of erosion with deep cartilage matrix loss and focal areas of exposure of the subchondral bone (Figure 6A, B); loss of proteoglycan staining, chondrocyte loss (Figure 6B, C) or chondrocyte clustering and hypertrophy and fissures of the cartilage (Figure 6B, C). Cartilage collapse into the epiphysis was also extensively observed (Figure 6C) and osteophytes were also found; some signs of synovial inflammation could also be observed (Figure 5E, F), although much less pronounced than at one and two weeks.

All measurements are presented in Figure 7. In control animals, changes were barely observed, being shown only when they were measurable. In collagenase-injected rats, there was a continuous and dose-dependent progression of OA evidenced by a higher degree of joint destruction for the 500 U dose and throughout the time of OA development for all parameters scored. Overall, there was a higher loss of cartilage matrix after six weeks and with the 500 U dose, as shown by the CMLW parameter. CMLW at the surface (0% depth, Figure 7A) was significantly enhanced both at four (P <0.05) and six weeks (P <0.001), reaching an average of 69 ± 8% of the superficial cartilage showing matrix loss. Deeper areas of the cartilage were only affected significantly six weeks after the injection of 500 U collagenase, as indicated by CMLW at the midzone (50% depth, Figure 7B), with 26 ± 6% (P <0.001) of the MTP width losing at least 50% of its depth. Exposure of the subchondral bone was observed in some sections after six weeks with the 500 U dose (Figure 6), as reflected in the CMLW at the tidemark (100% depth, Figure 7C) that showed that an average of 7 ± 3% of the cartilage width was lost. The TCDW measurement (Figure 7D) showed that cartilage degeneration, independent of the loss of the extracellular matrix, was already observable after one week for both doses, but continuously developed until the sixth week, when 84 ± 7% (P <0.001) of the cartilage width of animals injected with 500 U collagenase was affected by changes that included loss of proteoglycan staining, chondrocyte clustering and hypertrophy or chondrocyte death, and fissures (Figure 6). SCDW, which measures the areas where at least 50% of the cartilage depth is affected by any type of degenerative change (Figure 7E), showed the increased depth of cartilage alterations throughout time, being significant only at week 6 for the 500 U dose, when 42 ± 5% of its width was affected (P <0.05). This was further reinforced by the ZDR of lesions (Figure 7G-I), which showed that the average ratio of the cartilage depth affected significantly increased at six weeks. This parameter also showed that there is a different degree of destruction in different load bearing areas. In fact, the ZDR of lesions was higher in zone 2 (Figure 7H), where significant differences were already observed at week 1 for the 500 U dose (P <0.01), further increasing to 0.79 ± 0.06 at week 6 (P <0.001). Zone 3, next to the cruciate ligaments, showed a ZDR of 0.56 ± 0.09 at week 6 (P <0.01, Figure 7I), whereas in zone 1, the most medial area of the MTP, the ZDR was 0.41 ± 0.06 at week 6 (P <0.05, Figure 7G). The CDS, which takes into account the percentage of cartilage affected by any type of change, again showed the different degree of destruction at different load bearing areas (Figure 7J-L). Changes were significant only after six weeks and only for the 500 U dose of collagenase, and most pronounced in zone 2, where the average OA score was 2.9 ± 0.25 (out of 5, P <0.05, Figure 7K). The development of osteophytes was observable for both doses from week 4 onwards and was more pronounced six weeks after 500 U collagenase injection, when some large osteophytes could be observed (Figure 7F). Synovial inflammation (Figures 5 and 7M) was observed in control animals at one and two weeks, probably as a reaction to the intra-articular injection of saline. Collagenase-injected animals, however, showed a much higher degree of synovial inflammation at week 1, particularly for the 500 U dose, which diminished throughout time. No differences were observed in the medial joint capsule (Figure 7N). The growth plate thickness diminished over time, similarly to what was observed in controls (Figure 7O).

The correlation analysis between the behavioural data and the histopathological score as well as the knee diameter, for both doses, at weeks 1 and 6, is shown in Table 1. This correlation was performed with the data from the animals sacrificed at each time-point (n = 6/group). Noteworthy results include the significant correlation observed between the Knee-Bend score and the knee diameter (P <0.01) one week after injection for both doses. The CatWalk test did not correlate with the knee diameter. Some correlations were found for the Knee-Bend in the 250 U dose at both time points, namely with the TCDW and the ZDR for zones 2 and 3. The CatWalk showed no correlations for this dose at both time-points. For the 500 U dose, a larger number of correlations could be observed. At week 1, a correlation between both tests and the CMLW 0% was observed, most significant for the Knee-Bend test (P <0.001). This test also correlated with the TCDW (P <0.01), the ZDR at zone 2 (P <0.01) and with the synovial inflammation (P <0.05). At week 6, there were significant correlations between the Knee-Bend score and most of the histopathological parameters analysed (Table 1), namely with the CMLW at 0% depth (P <0.01), the TCDW (P <0.01), the SCDW (P <0.01), the ZDR at zones 2 (P <0.01) and 3 (P <0.05) and the CDS at zones 1 (P <0.01) and 2 (P <0.01) and the SMIS (P <0.01). CatWalk data also significantly correlated with the CMLW at 0% depth (P <0.05), the TCDW (P <0.05) and the CDS at zone 2 (P <0.05).

Expression of the ion channel TRPV1 was observed in small-sized neuronal cell bodies (Figure 8A, B), as is described for this receptor [39]. In control animals one week after saline injection, 29.4 ± 2.4% of DRG neurons expressed TRPV1 (Figure 8C). At the same time-point, animals injected with 500 U collagenase expressed TRPV1 in 31.9 ± 1.8% of DRG neurons (Figure 8C). Six weeks after saline injection, 25.0 ± 1.4% of DRG neurons were immunoreactive for TRPV1. In contrast, collagenase injected rats showed a significantly increased expression of TRPV1, with a value of 37.1 ± 2.2% positively labelled neurons (P <0.01, Figure 8C). Although TRPV1 expression in 500 U collagenase-injected rats at one week was not significantly different from controls at the same time-point, a significant difference was obtained from controls at six weeks (P <0.05).