Assessment of Osteoarthritis Functional Outcomes and Intra-Articular Injection Volume in the Rat Anterior Cruciate Ligament Transection Model

Abstract

The rat surgical anterior cruciate ligament transection (ACLT) model is commonly used to investigate intra-articular osteoarthritis (OA) therapies, and histological assessment is often the primary outcome measure. However, histological changes do not always correlate well with clinical outcomes. Therefore, this study evaluated functional outcomes in the rat surgical ACLT model and compared intra-articular injection volumes ranging from 20 to 50 μL. Unilateral ACLT was surgically induced, and static weight-bearing, mechanical allodynia, motor function, and gait were assessed in 4 groups of male, Sprague-Dawley rats (n = 6 per group). Intra-articular injections of 20 μL Dulbecco’s phosphate-buffered saline (DPBS), 50 μL DPBS or 50 μL of synthetic biomimetic boundary lubricant were administered once weekly for 3 weeks post-operatively. Structural changes were evaluated histologically at 20 weeks. Rat cadaver knees were injected with 20, 30, 40, or 50 μl of gadolinium solutions and were imaged using magnetic resonance imaging (MRI). Static weight-bearing, mechanical allodynia, and gait parameters in ACLT groups revealed differences from baseline and naïve controls for 4 weeks post-ACLT; however, these differences did not persist beyond 6 weeks. Different intra-articular DPBS injection volumes did not result in functional or histological changes; however, peri-articular leakage was documented via MRI following 50, 40, and 30 μl but not 20 μl gadolinium injections.

Statement of clinical significance:

Differences in functional parameters were predominantly restricted to early, postoperative changes in the rat surgical ACLT model despite evidence of moderate histologic OA at 20 weeks. Injection volumes of 20–30 μl are more appropriate for investigating intra-articular therapies in the rat knee.

Introduction

Anterior cruciate ligament (ACL) injury is a major risk factor for post-traumatic knee osteoarthritis (OA), with 50% of patients developing OA 10 to 20 years after injury¹. Despite tremendous efforts, there are still no disease-modifying treatments that effectively delay OA progression². Intra-articular administration of corticosteroids and hyaluronic acid have been used commonly, with autologous blood products, stem cells, bio-lubricants, and anti-inflammatory mediators being explored in recent years³. Rodent models are popular in early drug development and testing due to their low cost and rapid disease progression^4,5, and surgical induction methods are commonly selected to induce post-traumatic osteoarthritis (PTOA). However, to the authors’ knowledge, there is no data describing the physiological synovial fluid volume in the rat knee from which to infer an appropriate intra-articular injection volume for therapeutic testing. In the rat knee, 50 μL is a commonly reported injection volume^6–10, with some studies reporting volumes as great as 100 μL¹¹.

Functional outcome measurements are increasingly being investigated in rodent OA studies^12–14, as histological evaluation of joint structural changes does not always correlate well with functional impairment in animal models^12,15. Furthermore, human patients seek treatments that alleviate pain and restore function^16–18. However, it is still unclear what assessments are most sensitive, specific, and reliable for each animal model. Von Frey analgesiometry can be used to detect mechanical allodynia, or aversion to non-painful mechanical stimuli, by applying an increasing force to the hind paw¹⁹. Mechanical allodynia has been reported in surgical ACL transection (ACLT) with partial medial meniscectomy (MMx) and intra-articular injection of monoiodoacetate in rats²⁰, and in destabilization of the medial meniscus in mice²¹. Non-evoked pain can be measured by static weight-bearing using an incapacitance apparatus to detect decreased unilateral weight-bearing in rodent PTOA models^{11,19,22–26}. Rotarod analysis has been used to assess motor coordination in rodents by quantifying the duration of time that animals can remain on a rotating rod increasing in velocity^27,28. Gait analysis can be used to monitor the progression of PTOA and is commonly employed in humans, rodents and other pre-clinical models^29–33.

Therefore, the objectives of this study were to: (1) characterize static weight-bearing, mechanical allodynia, motor coordination, and gait changes in rats undergoing surgical ACLT following treatment with intra-articular injections from 0 to 20 weeks post-ACLT and (2) to determine the physiological joint volume of the adult knee in male, Sprague-Dawley rats, and to compare intra-articular injection volumes ranging from 20 to 50 μL. The ultimate goal of this study was to provide recommendations for intra-articular therapeutic testing in the rat surgical ACLT model.

Methods

Animals

All procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee of Cornell University. Male, 7- to 8-week-old Sprague-Dawley rats were housed in pairs under a standard 12hr light/dark cycle starting at 6 AM. The animals were allowed to move freely in their cages, fed a standard commercial diet, and were allowed access to tap water.

Study design

Before assigning individual rats to experimental treatment groups, baseline measurements for each functional or pain-related outcome were obtained. The animals were assigned to one of four groups (n = 6) using a random number generator and blinded to all investigators except for one (YW): naïve; ACLT with 20 μL injection of Dulbecco’s phosphate-buffered saline (DPBS); ACLT with 50 μL injection of DPBS; and ACLT with injection of 50 μL synthetic boundary bio-lubricant (sBBL) injection group (Fig. 1A).

Figure 1.

(A) Group assignment and (B) study timeline. A total of three intra-articular injections were performed beginning 1-week post-anterior cruciate ligament transection (ACLT), concurrent with weekly functional tests. Rotarod testing was performed on Day 1 after injection, von Frey testing was performed on Day 2 after injection, and gait and incapacitance testing were performed on Day 3 after injection. The order and timing of the functional testing remained the same after the injections were concluded.

ACLT was performed on the left knee via an open surgical approach under 2.5% isoflurane anesthesia at 1 L/min O₂, similar to previously described, at 10–12 weeks of age³⁴. Complete transection of the ACL was confirmed by a cranial drawer test. A daily dose of 4 mg/kg of ketoprofen was given subcutaneously for three days post-operatively. The animals could ambulate, eat, and drink ad libitum after surgery. Beginning one-week post-ACLT, intra-articular injections via a lateral parapatellar approach were conducted once weekly for a total of three weeks. The naïve group did not receive any injections. Weekly functional testing was initiated one-week post-ACLT, concurrent with the intra-articular injections. Upon completion of intra-articular injections at 4 weeks post-ACLT, functional testing was conducted every two weeks until completion of the study on week 20. The rotarod motor coordination test was performed on day 1 after injection, mechanical allodynia using a mechanical von Frey apparatus was performed on day 2 after injection, and static-weight bearing and gait testing were performed on day 3 after injection. The order and timing of the functional tests remained the same after the injections were concluded. At week 20, animals were euthanized by CO₂ followed by thoracotomy (Fig. 1B). All procedures were conducted in separate chambers from the housing area.

Synthetic boundary bio-lubricant (sBBL)

An sBBL mimicking the bottlebrush structure of lubricin was made by grafting hydrophilic polyethylene glycol (PEG) brushes to a polyacrylic acid (pAA) core. The pAA-g-PEG copolymer was synthesized via conjugation chemistry using 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methyl morpholinium chloride as the coupling agent based on a procedure developed previously³⁵. The final structure of the sBBL was pAA(75kDa)-PEG(5kDa)-NH.

Tissue processing and histology

Immediately following euthanasia, the left knees were fixed in 1× phosphate-buffered saline containing 4% paraformaldehyde and 1% cetylpyridinium chloride, pH 7.4, for 3 days, with the femur and tibia at an angle of ~120°. The samples were then decalcified in 10% ethylenediaminetetraacetic acid for 21–24 days until complete decalcification was confirmed radiographically (Faxitron, 60s exposure with 37V). Parasagittal sections (5 μm) of both medial and lateral femorotibial joints were obtained and stained with safranin O/fast green and hematoxylin and eosin. Cartilage degradation of both medial and lateral tibial plateaus and femoral condyles was evaluated using a modified Mankin system on a scale of 0–14 points scored by three independent observers (H. L. R., Y. W., and K. J. C.)³⁶; the median grade between the three observers was used for statistical analysis.

Static weight-bearing

Static weight-bearing was measured using an incapacitance meter (Cat #: 600MR, IITC Life Science). The average weight on each transducer was recorded over a 5 s interval, and the percent weight distribution on the left limb was reported.

Mechanical allodynia

Mechanical allodynia was assessed using an electric von Frey apparatus (Cat #: 2391, IITC Life Science). A 15 min acclimation period to the plexiglass chambers was allotted before testing. The paw withdrawal threshold (PWT), in grams, was recorded as the maximum force applied before a paw withdrawal was observed.

Motor coordination

Balance and motor coordination were assessed with forward movement on a rotarod (Cat #: 755, IITC Life Science). Two training sessions were conducted on two separate days within one week before ACLT, where animals were trained on a rotarod at a constant 5 rpm until they successfully balanced on the rod for 5 min for three consecutive trials. The animals were placed on a still rod before accelerating from 4 to 40 rpm over 5 min, and the duration of time in which the animals remained on the rod was recorded in seconds. The trials in which the animals actively jumped off of the front of the rod were excluded.

Gait analysis

Before ACLT, the animals were trained to walk through a transparent plexiglass walking track (27” in length) with a mirror oriented at 45° below the track to record foot placement before baseline gait recordings were obtained³⁷. Gait analysis was performed using commercially available motion tracking software (SIMI Reality Motion Systems GmbH). The speed, maximum vertical displacement, vertical asymmetry, print lengths, toe spread, intermediate toe spread, left step length, stride length, step length asymmetry, duty factor, and duty factor imbalance were calculated using custom MATLAB code.

Vertical displacement analysis

The position of the center of the last rib on the left side was tracked automatically using the SIMI motion tracking system and was used to represent the animal’s vertical position of the axial skeleton. The maximum vertical displacement was obtained by finding the range between the highest and lowest vertical positions throughout the entire trial (Eq. 1). Within each stride, there were two vertical peaks, representing upward vertical movements of the left rib corresponding with the left and right hindlimb push off. Each stride started at the left foot-strike position, and the left and right peaks of the vertical positions were calculated from the minimum vertical position within each stride. A value of 0 was assigned to a peak that was too small to be detected (Fig. 3A,B). The ratio of the left/right peak was calculated and was denoted as vertical asymmetry, where a ratio approximately equal to 1 indicates a symmetrical gait (Eq. 2).

Figure 3.

Schematics and calculations for the maximum vertical displacement and vertical asymmetry measurements; shaded area indicates one gait cycle for (A) naïve and (B) ACLT-operated rats. (C) Maximum vertical displacement and (D) vertical asymmetry for three groups of rats measured up to 20 weeks post-ACLT. N = 6, data reported as mean ± SEM. *p < 0.05 between DPBS and naïve, †p < 0.05 between sBBL and naïve, ‡p < 0.05 between DPBS and sBBL, n.s. not significant; generalized linear model with post​-hoc Tukey’s honestly significant difference (HSD), α = 0.05. The 20 μL DPBS group was omitted for clarity (Figure S3).

$$Maxverticaldisplacement=maxverticalposition-minverticalposition$$ Eq. 1

$$verticalasymmetry=\left(leftpeakvalue-baselinevalue\right)/\left(rightpeakvalue-baselinevalue\right)$$ Eq. 2

Gadolinium contrast magnetic resonance imaging (MRI) of rat cadaver knees

Thirty-six, disarticulated, frozen-thawed hindlimbs from 6- to 7-month-old, unoperated, male Sprague-Dawley rats were imaged using a clinical 3T MRI (Premiere DVMR 750, GE Healthcare) and a 14-channel flex coil (GE Healthcare), and protocols optimized for small anatomy (Table S1). The hindlimbs were stored at −20°C for up to 1 year before MRI imaging. A stock of 1:200 diluted gadolinium contrast with blue food coloring was made and each knee received a different volume slowly injected over 10–20 seconds using a 27g/0.5 mL syringe via the cranial margin of the patellar ligament: 50 μL (n = 12), 40 μL (n = 8), 30 μL (n = 8), and 20 μL (n = 8). Photographs of the visible leakage were obtained (Figure 5E–H) and limbs were immediately scanned (Figure 5A–D). The synovial fluid was manually segmented using ITKSnap, an open-source software package, and the fluid volume in the joint space was calculated pre- and post-injection of contrast agent. The post-injection T1 weighted images were scored for contrast leakage using a qualitative grading system of none (0), mild (1), moderate (2), and severe (3) at 5 locations around the knee (Fig. 6). The grades from all locations were summed to produce a total leakage score range of 0–15. In addition, joint distension was scored as follows: good (0), fair (1), or poor (2). All knees, regardless of the initial volume injected, scored as having “good” joint filling were segmented to calculate joint volume. All joints were segmented pre-contrast to obtain an average joint volume among the sample set.

Figure 5.

Sagittal plane T1W fat saturated post intra-articular gadolinium injection of a rat knee A) without and with B) mild, C) moderate, and D) severe contrast leakage at the caudal margin of the joint. Arrows depict leakage. Photographs of E) no, F) mild, G) moderate, and H) severe leakage along the cranial injection site. The use of blue dye improved conspicuity of the leakage on the in situ specimens.

Figure 6.

A, B, and D) sagittal plane and C) dorsal-oblique T1W FS MR images of the rat knee post intra-articular contrast demonstrating A) pre-patellar, B) caudal joint pouch, C) craniolateral tibial, and D) tibia-fibular contract leakage (arrowheads) as the four most common leakage sites. The 3D image includes labels corresponding to the regions from A-D and shows (in pink) the normal distribution of joint fluid.

Statistical analysis

A one-way analysis of variance was performed for animal weight at each time point and all functional data at baseline to assess whether there was any discrimination of study group assignment. The distributions of all data sets were tested using graphical methods with a quantile-quantile plot against the standard normal distribution. All functional data were analyzed using a linear mixed effects model and post hoc analysis was performed using least mean square Tukey honestly significant difference (HSD) test in the case of significant fixed effects. The analysis was carried out in the early-phase (weeks 1–4 post-ACLT), mid-phase (weeks 6–12 post-ACLT), late-phase (weeks 14–20 post-ACLT), and overall (weeks 1–20 post-ACLT). Tukey’s HSD test was used to assess the differences in modified Mankin grades between treatment groups. All statistics were performed using JMP Pro 14.0 software (SAS Institute Inc.). Significance was set at α = 0.05.

Results

Body weight did not differ between groups

The animals weighed 300.3 ± 5.0 g at the time of surgery. No differences in body weight were present between groups at any time point during the study and, at the end of the study, the body weights were 485.0 ± 12.0 g, 478.7 ± 15.0 g, 495.5 ± 8.0 g, and 451.2 ± 10.0 g for naïve (n = 6), ACLT with 20 μL DPBS injection (n = 6), ACLT with 50 μL DPBS injection (n = 6), and ACLT with 50 μL sBBL injection (n = 6) groups, respectively (mean ± SEM).

Rat knees showed PTOA structural changes 20 weeks post-ACLT

At week 20 post-ACLT, the articular cartilage surface of all naïve animals was smooth and free of cartilage matrix staining loss, and all naïve animals received mean total Mankin grades of 0 (Fig. 4). All ACLT groups, regardless of intra-articular treatment, showed articular cartilage histopathological changes, including thickened tibial cartilage and chondrocyte clustering. Decreased cartilage surface proteoglycan staining and mild surface fibrillation were observed in all three ACLT groups (Fig. 4A, empty and black arrowheads, respectively). Hypertrophic chondrocytes and irregular endochondral ossification were also observed in all there ACLT groups (examples shown in Fig. 4A, yellow arrows and arrowheads, respectively). The Mankin grades for both 20 μL and 50 μL DPBS-injected groups were greater as compared to naïve (p = 0.02 and p = 0.01, respectively), whereas the Mankin grade of the sBBL injection group was similar to the naïve group (p = 0.23) (Fig. 4B).

Figure 4.

A) Representative SafO/Fast Green staining of the left knee. Top panel, low magnification (4x), scale bar = 400 μm; bottom panel, high magnification (16x) of areas indicated with dotted line, scale bar = 100 μm. Empty arrowheads: proteoglycan loss; yellow arrows: hypertrophic chondrocytes; black arrowheads: cartilage surface fibrillation; black arrows: fissure; yellow arrowheads: irregular endochondral ossification. Black bracket on the femoral side in the anterior cruciate ligament transection (ACLT) + 50 μL synthetic boundary bio-lubricant (sBBL) group indicates a sectioning artifact. (B) Box and whiskers plots of modified Mankin grade of the medial tibial plateau, representing the median, maximum, minimum, and quantiles. N = 6, letters represent different levels in the Tukey’s honestly significant difference (HSD) test, α = 0.05. F, femur; T, tibia.

ACLT animals exhibited reduced weight-bearing and increased mechanical allodynia in the early phase post-ACLT

The ACLT surgery resulted in static weight-bearing imbalance and mechanical allodynia in the early-phase in both DPBS and sBBL groups as compared to naïve rats. The sBBL group demonstrated more pronounced static weight-bearing imbalance as compared to the 50 μL DPBS-treated group only in the early phase (Fig. 2A). Static weight-bearing distributions and PWTs for both 50 μL DPBS- and sBBL-treated ACLT groups returned to similar levels as those of naïve controls in the mid-phase and did not differ from naïve controls in the late-phase (Fig. 2A,B).

Figure 2.

A) Static weight-bearing measured by incapacitance meter, reported as % weight on the left side, B) mechanical allodynia measured by electronic von Frey apparatus, reported as left paw withdrawal threshold (PWT) in grams, and C) motor coordination measured by endurance on rotarod, reported as time in seconds, for three groups of rats measured up to 20 weeks post-ACLT. Dotted line in panel (A) indicates an even, or 50% weight distribution; dotted line in panel (B) represents the average baseline PWT of all three groups. A decrease in these measurements indicates a more severe lameness (A) or increased pain sensitivity (B). N = 6, data reported as mean ± SEM. *p < 0.05 between DPBS and naïve, †p < 0.05 between sBBL and naïve, ‡p < 0.05 between DPBS and sBBL, n.s. not significant; generalized linear model with post-hoc Tukey’s honestly significant difference (HSD), α = 0.05. The 20 uL DPBS group was omitted for clarity (Figure S3).

ACLT animals did not demonstrate changes in motor coordination or balance

The average time that animals remained on the rod increased in the first few weeks after weekly training and testing, suggesting an improved performance by learning (Fig. 2C). No differences in rotarod performance were observed between naïve and 50 μL-injected groups at any-phase post-ACLT, and significant inter-individual variability was noted.

ACLT animals exhibited increased gait asymmetry in the early phase post-ACLT

No differences in speed were observed between naïve and 50 μL-injected groups at any phase (Supp. Fig. 1A). Changes in gait type were observed in groups that received ACLT surgeries in the early phase, with lame animals demonstrating a gait resembling a three-beat canter. Both the left print length and toe spread in the ACLT + sBBL injection group decreased as compared to naïve rats in the early-phase, with no differences between any groups in the mid- and late-phases (Supp. Fig. 2). The ACLT + sBBL injection induced a decrease in the left stride length in both early- and mid- phases, whereas the left stride length of the ACLT + 50 μL DPBS injected group remained comparable to that of naïve controls (Fig S1B). The ACLT + sBBL injection group also showed increased step length asymmetry in the early-phase compared to naïve controls (Supp. Fig. 1C). Duty factor imbalance was more pronounced in the ACLT + sBBL group as compared to the ACLT + 50 μL DPBS group in the early-phase, whereas both groups showed increased duty factor imbalance compared to naïve controls (Supp. Fig. 1D). Negative values for the duty factor imbalance resulted from the decreased left hindlimb stance time, another indicator of compensation by the right hindlimb.

ACLT animals exhibited a greater maximum vertical displacement in early and late phases post-ACLT

In the early-phase after surgery, increased maximum vertical displacement and vertical asymmetry were observed for the ACLT + 50 μl DPBS group as compared to the naïve group, indicative of an asymmetric gait resulting from the left hindlimb lameness. Both maximum vertical displacement and vertical asymmetry returned to similar levels as naïve controls in the mid-phases (Fig. 3C,D). The ACLT + sBBL group had a greater maximum vertical displacement as compared to the naïve group in the late-phase and for Weeks 1–20 (Fig. 3C), and greater vertical asymmetry was noted for the ACLT + sBBL group as compared to the naïve group for Weeks 1–20 (Fig. 3D).

Functional performance and structural changes did not differ between 20 and 50 μL intra-articular injections post-ACLT

Both ACLT + 20μL DPBS and ACLT + 50μL DPBS groups showed similar functional and pain-related outcomes, wherein the early-phase the rats demonstrated reduced static weight-bearing, reduced PWTs and increased gait asymmetry which recovered to levels similar to the naïve control group in the mid-phase (Supp. Fig. 3). No differences in functional parameters were observed between the ACLT with 20 μL DPBS and ACLT with 50 μL DPBS injection groups at any phase (Supp. Fig. 3). Histological analysis demonstrated similar PTOA structural changes in both ACLT with 20 μL DPBS and ACLT with 50 μL DPBS injection groups 20 weeks after ACLT, with no differences in the Mankin grades observed between these two groups (Fig. 4B).

Greater intra-articular injection volume caused peri-articular leakage

The average pre-injection joint volume for rat knees was 7.1 ± 0.2 μL (mean ± SEM) with a range of 5.6 to 10.4 μL and a median of 7.6 μL. Following injection, the volume of fluid in knees scored as having “good” contrast filling displayed an average volume of 24.2 ± 1.2 μL with a range of 19.0 to 31.2 μL. No differences in joint fluid volume were observed between the 50 and 20 μL groups post-injection (Supplemental Table 2). In rats receiving 50 μL DPBS and sBBL injections for the in vivo study, peri-articular leakage could be appreciated due to subcutaneous accumulation of injectate immediately post-injection. Similar to observations in live animals, contrast-enhanced MRI revealed peri-articular leakage in all injection groups but was greatest in the 50 μL injection groups (Supp. Fig. 4).

Discussion

Here, we have shown that static weight-bearing, mechanical allodynia, and vertical gait asymmetry detect early changes and recovery in the rat surgical ACLT model. The new vertical displacement and gait asymmetry parameters described provide a rapid and objective assessment of gait abnormalities in response to ACLT that could provide more rapid, high throughput analyses. However, no long-term functional or pain-related behaviors associated with chronic PTOA were observed in this study, even at 20 weeks post-ACLT, despite histological PTOA structural changes. In addition, we have documented reproducible peri-articular leakage following high-volume intra-articular injections of contrast agent into the rat knee via MRI, suggesting that injection volumes of 20–30 μL may be most appropriate for the adult, male Sprague-Dawley physiological knee volume of ~7.1 μL.

This is the first study to utilize the vertical rib position to assess gait asymmetry in rats. A wireless inertial sensor system has been developed to objectively detect lameness in horses by assessing vertical movements of the head and pelvis while trotting over ground³⁸. We took a similar approached by analyzing the change of vertical position (presented by the tattoo mark in the middle of the last rib on the left flank) in rats. We found a significant increase in the maximum vertical displacement in the early-phase post-ACLT, whereas the vertical displacement for naïve animals remained consistent throughout study. After analyzing the trace of the vertical position, we found that there are two distinct peaks in each stride, sandwiched between the foot-strike and toe-off positions of each hindlimb. By taking the ratio of left and right peaks, we were able to capture the vertical displacement changes caused by gait asymmetry within each stride. These peaks are not to be confused with those normally described in vertical ground reaction force analysis, where the peaks in ground reaction forces occur near the foot strike position^32,39.

Vertical asymmetry showed a significant increase in the early phase post-ACLT, similar to static weight-bearing, mechanical allodynia, and duty factor imbalance. Because of the normalization to the baseline within each stride, inter-stride variability is accounted for in this parameter. Together, maximum vertical displacement and vertical asymmetry can reflect both lameness of the operated side and shuffling behavior in the form of bilateral compensation of a unilateral injury, similar to what was observed in the rat medial meniscal transection model³². In additionally, they are easily extracted from the trace of a single point, without manual digitization of toe-off and foot-strike positions, thus minimizing the variation due to subjective differences and allowing for rapid batch processing.

Static weight-bearing measured via incapacitance meter reflected post-surgical unloading followed by recovery but showed no changes related to PTOA in the late-phase. The initial reduction in weight-bearing is well-documented in rodent models^11,23–26; however, it has been hypothesized that the rats can offload weight from the hindlimbs to forelimbs in static weight-bearing. The presence of mechanical allodynia for up to 4 weeks post-ACLT is likely dominated by nociceptive joint pain due to surgery rather than the effect of PTOA, in agreement with previous findings where the reduction and recovery in PWT was similar for both sham-operated and ACLT rats²². Conversely, Silva et al observed a decrease in PWT post-ACLT from 10 weeks until study completion at 12 weeks²².

Similar to previous studies, there were no differences in rotarod times between naïve and operated animals, although a trend for an improved performance after repeated tests was observed^27,40. These findings highlight some of the disadvantages of using the rotarod test to identify motor deficiency caused by surgical ACLT, including significant inter-individual variability and the requirement for extensive training before to study initiation. However, out of all the functional tests, the rotarod was the only test with a trend for distinguishing between the ACLT + intra-articular injection and naïve groups in the late phase of the study. In retrospect, the two training sessions allocated to rotarod training in this study was likely insufficient. With more subject numbers and intensive training, rotarod testing might be a valuable parameter to assess the changes in motor coordination or pain related to longer-term PTOA functional deficits.

To our knowledge, no prior studies have evaluated multiple functional parameters in the surgical rat ACLT model as late as 20 weeks post-operatively. However, a biphasic response was observed in a rat ACLT + MMx model with increased pain observed in the initial 5 weeks and around 10 weeks post-surgery²⁷. Although synovitis and decreased functional performance were observed in another surgical rat ACLT model as early as 30-days post-operatively, the paw print area and PWT differences between operated and control animals started to converge at 60 days post-ACLT⁴¹. Thus, a more severe model, such as ACLT + MMx, might be needed to observe late-phase functional differences arising from PTOA. Variation between studies may also be related to variation in the severity of injury induced by different surgeons.

A limitation of the vertical asymmetry analysis is that the marker was placed on the left rib rather than on the dorsal midline. This is a limitation due to the 2-dimensional nature of the motion tracking software and the requirement for the marker to be placed in a location visible to the video camera; however, given that the vertical asymmetry measurement was sensitive, straightforward, and high-throughput, it might be possible to design a pelvic or spine-mounted marker that could be used for videographic motion analysis. Alternatively, inertial sensor systems that could be secured to the dorsal midline of rodents could have potential value for both walking track analysis and free-cage data collection.

The sBBL injection was included as a potential viscosupplementation therapy to ameliorate the progression of PTOA. Although no functional nor pain-related differences were found 20-weeks post-ACLT, the histopathological analysis demonstrated joint structural changes. The sBBL-treated group revealed slight improvements in histopathology as compared to DPBS-treated groups, though these differences were not statistically significant.

Limitations of this study design included the use of only a naïve negative control rather than both naïve and sham-operated negative controls. In addition, an analgesic medication, such as buprenorphine, would be informative as a positive control for systemic analgesia; however, the duration of activity of intra-articular analgesics are limited. If an analgesic medication were to be included in a future study, its short-term effect should be taken into consideration. Another limitation of this study is that the parasagittal sectioning plane used for histopathological analysis prohibited assessment of osteophyte formation, which is commonly associated with OA development in rats⁴². With respect to the intra-articular injection volume study, we cannot rule out of the fact that cadaver tissues may be more friable than tissues in live rats, with an increased likelihood of leakage due to tearing of the joint capsule.

One prior study investigated intra-articular injection volumes in rat knees, where intra-articular injections exceeding 30 μL commonly resulted in subdermal leakage outside of the joint capsule⁴³. It is important that injection volumes in pre-clinical animal models are physiological or approximately equivalent to injection volumes delivered to human patients, especially when pain and functional outcomes are being measured. Typical intra-articular injection volumes for the human knee do not exceed 2 mL and one study examined 61 patients who underwent endoprosthetic joint replacement and found their knee joint volumes range from 40 to 290 mL with an average of 131 mL^44,45. Although joint volume may not scale linearly with body weight, our estimates, in vivo experience, and cadaveric MRI imaging studies all suggest that 50 μL is an excessive volume for the rat knee. Contrast agent leakage was observed in the caudal pouch, pre-patellar, pre-tibial, and tibial-fibular areas following intra-articular injections and the leakage was pronounced in 50 μL injections. Although 10–15 μl may be more ideal, adult, male Sprague-Dawley knee joints tolerated 20 μl injections without substantial peri-articular leakage. No differences in functional performance were observed between 20 and 50 μL DPBS treated groups; however, the impact of peri-articular leakage may be more significant with therapeutic or high molecular weight injections as compared to placebo injections.

Together, our findings will inform the design of future rodent studies investigating the surgical ACLT model, including selection of relevant outcome measures and injection volumes. In addition, these data suggest that histological evidence of PTOA at the study termination does not necessarily coincide with pain- or functional derangements in rodent models, further motivating evaluation of functional parameters for investigation of OA therapeutics in combination with structural analyses.

Extended Methods

Animals

All procedures were performed according to protocols approved by the Institutional Animal Care and Use Committee of Cornell University (Protocol #2017-0084). Male Sprague-Dawley rats were purchased from Harlan Sprague-Dawley, Inc. at 7–8 weeks of age and were housed in pairs under a standard 12hr light/dark cycle starting at 6 AM. The animals were distributed randomly into cages in the same room, independent of treatment group. The animals were allowed to move freely in their cages, fed a standard commercial diet, and allowed access to tap water. The animals were given a minimum of 3 days of acclimation upon arrival, followed by earn notching for identification purposes.

Study design

Prior to assigning individual rats to experimental treatment groups, animals were tattooed for gait analysis motion tracking and were given a minimum of 3 days to recover prior to obtaining baseline measurements for each pain- or functional outcome. For motion-tracking gait analysis, the animals were trained to walk through a 27” long transparent plexiglass track on cue using similar techniques as previously described³⁹. Two training sessions were conducted where the completion was marked by three consecutive trails where the animals walked through the track on cue without stopping. For rotarod motor coordination testing, animals were trained to walk on a rotarod at a constant 5 rpm velocity until they successfully balanced on the rod for 5 min for three consecutive trials. Baseline data for all animals were then obtained, with details described in their respective sections below. The animals were assigned to one of four groups (n = 6) using a random number generator: a naïve, unoperated control; surgical ACLT with 20 μL injection of Dulbecco’s phosphate-buffered saline (DPBS); surgical ACLT with 50 μL injection of DPBS; and surgical ACLT with injection of 50 μL synthetic boundary bio-lubricant (sBBL) injection group (Fig. 1A).

All investigators were blinded to the treatment group assignment during all functional testing, with the exception of one investigator (YW). ACLT was performed on the left knee via an open surgical approach, similar to previously described, at 10–12 weeks of age⁴⁰. The surgery was performed under 2.5% isoflurane anesthesia with 1 L/min O₂. The left knee joint was exposed via a medial parapatellar approach, the patella was subluxated laterally, and the knee was fully flexed to expose the ACL to enable transection with a No. 11 scalpel blade. Complete transection of the ACL was confirmed immediately after transection by demonstration of cranial displacement of the tibia with respect to the femur, or a cranial drawer test. The patella was replaced, and the patellar tendon was sutured with 3–0 poly (ethylene terephthalate) in a simple, interrupted pattern, followed by skin closure with 4–0 nylon in a simple, interrupted pattern. A daily dose of 4 mg/kg of ketoprofen was given subcutaneously for three days post-operatively. The animals were allowed to ambulate, eat, and drink ad libitum after surgery. Beginning one-week post-ACLT, 50 μL of DPBS or sBBL or 20 μL of DPBS was injected intra-articularly via a lateral parapatellar approach once weekly for a total of three weeks. The naïve group did not receive any injections. Weekly functional testing was initiated one-week post-ACLT, concurrent with the intra-articular injections. The first functional testing started at least 24 hours post-injection in a suite separated from the housing room, and tests were performed in the following order: rotarod, mechanical allodynia, static weight-bearing, and gait analysis. Upon completion of intra-articular injections at 4 weeks post-ACLT, functional testing was conducted every two weeks until completion of the study on week 20. At week 20, animals were euthanized by CO₂ followed by thoracotomy (Fig. 1B).

Synthetic boundary bio-lubricant (sBBL)

A synthetic boundary bio-lubricant (sBBL) mimicking the bottlebrush structure of lubricin was made by grafting hydrophilic polyethylene glycol (PEG) brushes to a polyacrylic acid (pAA) core. The pAA-g-PEG copolymer was synthesized via conjugation chemistry using 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methyl morpholinium chloride (DMTMM) as the coupling agent based on a procedure developed previously⁴¹. The final structure of the sBBL was pAA(75kDa)-PEG(5kDa)-NH.

Tissue processing and histology

Immediately following euthanasia, the left knees were fixed in 1x PBS containing 4% paraformaldehyde and 1% cetylpyridinium chloride, pH 7.4, for 3 days, with the femur and tibia at an angle of approximately 120°. The samples were then decalcified in 10% ethylenediaminetetraacetic acid for 21–24 days until complete decalcification was confirmed radiographically (Faxitron, 60 s exposure with 37 V). After embedding in paraffin wax, 5 μm parasagittal sections of both medial and lateral femorotibial joints were obtained and stained with safranin O/fast green and hematoxylin and eosin (H&E). Cartilage degradation of both medial and lateral tibial plateaus and femoral condyles was evaluated using a modified Mankin system on a scale of 0–14 points by three independent observers (HLR, YW and KJC); the median grade between the three observers was used for statistical analysis.

Static weight-bearing

Static weight-bearing was measured using an incapacitance meter (IITC LIFE SCIENCE) as previously described⁴², which reports the weight distribution between ipsilateral (left) and contralateral (right) limbs in the stationary animal. Rats were trained to stand in the incapacitance meter holding chamber, with each hind paw in the center of each transducer. The average weight on each transducer was recorded over a 5 sec interval, and the percent weight distribution was calculated with the following equation (Eq. 1). Five trials were obtained during each assessment with the mean of the five trials used for analysis. Baseline meassuremnts were obtained within one week prior of ACLT surgery. Two assessments were conducted on separate days, and the average of the 4 assessments was recorded for baseline measurement.

$$\mathrm{\%}weightdistribution=\frac{weightofipsilaterallimb}{(weightonipsilaterallimb+weightoncontralaterallimb)}\times 100\mathrm{\%}$$ Eq. 1

Mechanical allodynia

Mechanical allodynia was assessed using an electric von Frey apparatus (IITC LIFE SCIENCE). Rats were placed into plexiglass chambers (IITC LIFE SCIENCE) with a mesh metal flooring, allowing the stimulus to be applied to the hind paws from below. A 15 min acclimation period to the plexiglass chambers was allotted prior to testing. A non-painful stimulus was applied to the center of the hind paws using the electric von Frey probe with a rigid tip attachment. The left hind paws were stimulated first, followed by the right hind paw for each trial. The paw withdrawal threshold (PWT), in grams, was recorded as the maximum force applied before a paw withdrawal was observed. Three trials were obtained during each assessment, with at least 15 min between trials. The mean of the three trials were used for analysis. The baseline measurements were conducted the same way within one week prior to the ACLT surgery.

Motor coordination

Balance and motor coordination were assessed with forward movement on a rotarod (IITC LIFE SCIENCE). Two training sessions were conducted as described earlier on two separate days followed by baseline acquisition which followed the same procedure and conducted within one week prior to ACLT surgery. The animals were placed on a still rod prior to accelerating from 4 to 40 rpm over 5 min, and the duration of time in which the animals remained on the rod was recorded in seconds. Three trials were obtained during each assessment, with at least 20min between trials. The trials in which the animals actively jumped off of the front of the rod were excluded. The mean of all qualified trials was used for analysis.

Gait analysis

Upon receiving the animals, black tattoo marks were placed on the center of the left last rib, left hip, left knee, left hock, center of the left-most metatarsal, and the center of each hind paw. This procedure was done under 2.5% isoflurane anesthesia supplemented with 1 L/min O₂, and meloxicam was administered at 5 mg/kg subcutaneously on and after the day of the procedure. A one-week recovery period was allotted prior to gait analysis baseline testing. The tips of the first, second, fourth and fifth digits of each hind paw were marked with a black marker (Sharpie) before each assessment. Prior to ACLT, the animals were trained to walk through a transparent plexiglass walking track with a mirror oriented at 45° to record foot placement before baseline gait recordings were obtained.³⁹ For each assessment, three videos of the animals walking along the track were recorded at 100 frames per second (fps). The middle 20” of the track was used for analysis. Gait analysis was performed using commercially available motion tracking software (SIMI Reality Motion Systems GmbH). The speed, maximum vertical displacement, vertical asymmetry, print lengths, toe spread, intermediate toe spread, left step length, stride length, step length asymmetry (Eq. 2), duty factor (Eq. 5) and duty factor imbalance (Eq. 4) were calculated using custom MATLAB code (supplemental information).

$$steplengthsymmetry=\frac{leftsteplength}{stridelegnth}$$ Eq. 2

$$dutyfactor=\frac{stancetimeofthelimb}{stridetimeofthelimb}$$ Eq. 3

$$dutyfactorimbalance=leftdutyfactor-rightdutyfactor$$ Eq. 4

Vertical displacement analysis

For gait analysis, the position of the center of the last rib on the left side was tracked automatically using the SIMI motion tracking system and used to represent the animal’s vertical position of the axial skeleton. The maximum vertical displacement was obtained by finding the range between the highest and lowest vertical positions throughout the entire trial (Eq. 5). Within each stride, there were two vertical peaks, representing upward vertical movements of the left rib corresponding with left and right hindlimb pushoff. Each stride started at the left foot-strike position, and the left and right peaks of the vertical positions were calculated from the minimum vertical position within each stride. A value of 0 was assigned to a peak that was too small to be detected (Fig. 3A, B). The ratio of left/right peak was calculated and was denoted as vertical asymmetry (Eq, 6), where a ratio approximately equal to 1 indicates a symmetrical gait.

$$maxverticaldisplacement=maxverticalposition-minverticalposition$$ Eq. 5

$$vertcialasymmetry=\frac{leftpeakvalue-baselinevalue}{rightpeakvalue-baselinevalue}$$ Eq. 6

Gadolinium contrast magnetic resonance imaging (MRI) of rat cadaver knees

Thirty-six, disarticulated, frozen-thawed limbs from 6–7 month-old, male Sprague-Dawley rats were imaged using a clinical 3T MRI (Premiere DVMR 750, GE Healthcare, Waukesha, WI) and a 14 channel flex coil (GE Healthcare, Waukesha, WI) and protocols optimized for small anatomy (Supplemental Table 1). A stock of 1:200 diluted gadolinium contrast with blue food coloring was made and each knee received a different volume slowly injected over 10–20 seconds using a 27g/0.5 mL syringe via the cranial margin of the patellar ligament: 50 μL (n=12), 40 μL (n=8), 30 μL (n=8), and 20 μL (n=8). Photographs of the visible leakage were obtained, and limbs were immediately scanned (Fig. 5). The synovial fluid was manually segmented using ITKSnap, an open source software package. The post-injection T1 weighted images were scored for contrast leakage using a qualitative grading system of none (0), mild (1), moderate (2), severe (3) at 5 locations around the knee (Fig. 6)

1. Superficial margin of the patellar tendon/ligament (cranial/proximal)

2. Superficial margin of the tibia (cranial/distal/medial)

3. Caudal to the knee joint (caudal)

4. Distal to the tib-fib articulation (distal/lateral)

5. Other (variable)

The grades from all locations were summed to produce a total leakage score range of 0–15. In addition, joint distension was scored as: good (0), fair (1), or poor (2).

0: good – smooth margins, evident distension

1: fair – some buckled margins, contrast present but more room for distension

2: poor – flattened joint, poor filling

All knees, regardless of initial volume injected, scored as having “good” joint filling were segmented to calculate joint volume. All joints were segmented pre-contrast to obtain an average joint volume among the sample set.

Statistical analysis

All investigators were blinded to subject treatment group allocations during histologic and functional analysis data acquisition, with the exception of YW. A one-way ANOVA was conducted for animal weight data at each time point. A one-way ANOVA was also performed for all functional data at baseline to assess whether there was any discrimination of study group assignment. The distributions of all data sets were tested using graphical methods with a quantile-quantile plot against the standard normal distribution. All data was analyzed using a linear mixed effects model, which included the group (naïve, ACLT + 50 μL DPBS injection, ACLT + 50 μL sBBL injection, and ACLT + 20 μL DPBS injection), week, and a group week interaction term. The animal ID was included as a random effect term, and post-hoc analysis was performed using least mean square Tukey HSD test in the case of significant fixed effects. The analysis was carried out in the early-phase (week 1–4 post-ACLT), mid-phase (week 6–12 post-ACLT), late-phase (week 14–20 post-ACLT), and overall (week 1–20 post-ACLT). Inter-rater agreement of the modified Mankin scores was calculated using the linear-weighted kappa statistic (MedCalc Software Ltd). Tukey’s HSD test was used to assess the differences in modified Mankin grades between treatment groups. All statistics were performed using JMP Pro 14.0 software (SAS Institute Inc.). Significance was set at α = 0.05.