Study design
The study was an experimental non-randomized case-control study. The study was approved by the Danish Animal Experiments Inspectorate (approval no. 2020-15-0201-00602 and 2017-15-0201-01314) as well as the local Ethical and Administrative body of the Department of Veterinary Clinical Sciences, University of Copenhagen (# 2020-016 and #2017-010). The horses were all purchased from a Standardbred sales and training stable.
OA was surgically induced using the carpal osteochondral fragment model (day − 18)24,48–55. Treadmill exercise was initiated on day − 4. Blood and synovial fluid (SF) were sampled on days − 18, -14, 0, 3, 10, 17, 24, 31, 38, 45 and 52. All baseline samples and registrations were obtained before surgery on day − 18. At every sampling time, clinical parameters (pain behavior, skin temperature, joint swelling, and lameness analysis) were measured before synovial sampling. The treatment group was treated with integrin α10-MSCs on day 0. The horses were euthanized on day 53 or 54 (Fig. 1). After study termination, the carpi were subjected to magnetic resonance imaging (MRI), computer tomography (CT), macroscopic evaluation and tissue sampling for histology and MSC-tracking through polymerase chain reaction (PCR) detection of the Y-chromosome of the male MSCs in the female recipients.
In order to evaluate effect of treatment, integrin α10-MSC treated horses were compared to an untreated group of horses with induced OA from a previous study, where the same OA model was used, and where data were collected using identical methods. Consequently, the parts of the study could not be randomized, and blinded observations were not possible for some parameters (lameness, pain behavior, carpal skin temperature, and joint circumference). Blinding was achieved for all postmortem analyses including histology, gross pathology, CT/MRI, and SF analyses. All horses were housed in the same facility, handled by the same group of people, and followed the same protocol.
Animals
Seventeen healthy Standardbred trotters aged 3 to 7 years (mean: 4.7 years; median 4 years), weight range 396 to 535 kg (mean: 472 kg; median 470kg), 15 mares and two geldings (all MSC treated horses were mares) were included in the study. There were eight horses (all mares) in the treated group and nine horses in the untreated group. The horses were free from visual lameness, had no response to carpal flexion test, did not have palpable joint effusion, presented no abnormalities on radiographic examination, and had no signs of synovial inflammation as judged by routine parameters (total protein concentration (TP), total nucleated cell count (TNCC) and differential cell count (DCC)) of the middle carpal joints. The horses were habituated to treadmill exercise before commencing the study.
Isolation And Selection Of α10-mscs
Equine MSCs were isolated from adipose tissue from a 7-year-old healthy gelding, culture expanded until P3 and integrin α10β1 expressing MSCs were selected as previously described20,42,56. Only integrin α10β1-selected MSCs (integrin α10-MSCs) were used in this study.
Osteoarthritis Model
Induction surgery
On study day − 18, osteoarthritis was surgically induced. Surgery was performed by European specialists in large animal surgery (European College of Veterinary Surgeons [ECVS] diplomates, SJ and CL). In all horses, the left carpus was the OA-joint and the right carpus served as the sham-operated (arthroscopy only) Control-joint. The carpal osteochondral fragment model used in this study is well established24, 48–55. In brief, an osteochondral fragment was created arthroscopically in the third facet of the radial carpal bone (RCB) in the OA joint (Fig. 2). Postoperative medication consisted of flunixin 1.1 mg/kg SID for a total of two days, and benzyl penicillin 22,000 IU/kg QID plus gentamicin 6.6 mg/kg SID for a total of three days. The horses were confined to stall rest for 14 days after surgery. The horses were allowed free pasture exercise every day after study day − 4.
Treadmill exercise program
Starting 14 days after OA induction (day − 4), the horses were exercised on the treadmill once a day five days a week using the following program: 2 min slow trot 16–19 km/h (4.4–5.3 m/s), 2 min fast trot 32 km/h (9 m/s), 2 min slow trot 16–19 km/h (4.4–5.3 m/s). Horses were walked by hand before and after trotting exercise to warm up and cool down.
MSC treatment
On study day 0 (18 days after OA induction surgery), horses in the treatment group (horses #1–8) were treated with 2 x 107 equine allogeneic adipose tissue-derived and integrin α10β1-selected male mesenchymal stem cells (integrin α10-MSCs) in 4 ml DMSO cryopreservation medium (Cryostor, BioLife Solutions). The MSCs were thawed in a sterile water bath with a constant temperature of 37°C, aspirated slowly in a sterile fashion through a 14G canula and injected IA with a 20G canula over a minimum of 10 seconds. Synovial fluid was sampled from both middle carpal joints before treatment. No antibiotics or NSAIDs were administered. The horses were stall rested for 3 days after treatment. Lameness at a walk, carpal skin temperature, joint circumference of the middle carpal joint and pain behavior were recorded daily. The untreated group (horses #10–17) were left untreated.
Clinical Assessment
On sampling days, skin temperature was measured on the dorsal aspect of the middle carpal joint with an infrared thermometer (Fluke 572) and the circumference of the middle carpal joint was measured with a measuring tape. The horses were assessed for lameness by an ECVS (European College of Veterinary Surgery) diplomate (STJ or CL) using the AAEP (American Association of Equine Practitioners) lameness scale scale57. Pain reaction to flexion of the carpi was assessed subjectively (graded as 0 to 3) by flexing the joint with moderate tension for 60 seconds before letting the horse trot58. The observers were not blinded to treatment group, but they were blinded to results from the other group during data collection. In addition, the horses were assessed for lameness objectively using a sensor-based movement analysis system (Lameness Locator® by Equinosis LLC).
Synovial Fluid Analysis
Synovial fluid sampling
Horses were allowed to rest for a minimum of two hours after the lameness assessment before collecting synovial fluid (SF) samples. A 21G needle was inserted in the middle carpal joint medial to the common digital extensor tendon and 4 mL of SF was aspirated. The SF was immediately divided in two aliquots: one EDTA coated tube kept at room temperature for standard SF analysis (TNCC, TP and DCC) and one EDTA coated tube kept on ice for preparation of supernatants for cytokine/chemokine analysis and cell pellet for MSC tracking. All samples were processed within one hour. TP was assessed by refractometry. TNCC was quantified using a hemocytometer. Differential cell counts were evaluated using cytospin analysis. SF for cytokine analyses and MSC tracking was centrifuged at 1000xG for 20 minutes at 4oC. Supernatants were transferred to Protein LoBind Eppendorf tubes and frozen at -80oC, while cell pellets were washed in PBS and snap frozen in liquid nitrogen. All samples were stored at -80oC until analysis.
Synovial fluid inflammatory mediators
SF cytokines and chemokines (fibroblast growth factor-2 (FGF2), eotaxin (CCL11), granulocyte colony stimulating factor (G-CSF), granulocyte monocyte colony stimulating factor (GM-CSF), fractalkine (CX3CL1), interleukins (IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8 (CXCL8), IL-10, IL-12 (p70), IL-13, IL-17A, IL-18), macrophage chemoattractant protein 1 (MCP-1), tumor necrosis factor alpha (TNF-α), interferon gamma (INY-γ), GRO (CXCL1), and RANTES (CCL5) were quantified using a commercially available equine multiplex assay (MILLIPLEX MAP Equine Cytokine/Chemokine Magnetic Bead Panel, EMD Millipore) according to manufacturer’s protocol. All samples were analyzed in duplicate using a 96-well platform. PGE2 (Enzo® Life Sciences) was analyzed by ELISA according to manufacturer’s protocol.
We had a full sample set available for every timepoint from the treated group (available: n = 88/88), whereas some samples were missing in the untreated group (available: n = 63/99). No samples from horse #9 (untreated) were available; on day 10 only 2/9 samples where available; and horse #13 was euthanized on day 31.
Postmortem Assessment And Sampling
On day 53 or 54 (71 or 72 days after OA induction) the horses were euthanized with an overdose of pentobarbital.
Diagnostic imaging
Immediately after euthanasia, the left (OA) carpi of all horses were subjected to MRI and CT (see parameters in Table S1 and Table S2, Additional file 1). Images were evaluated by an veterinary imaging specialist (American College of Veterinary Radiology [ACVR] diplomate (JFG) using a previously described semi-quantitative scoring system adapted for use with a lower field strength59. Image analysis was blinded to treatment group, and the CT and MRI studies were evaluated concurrently. Images were analyzed in Osirix MD (v.10.0.4, Pixmeo SARL). The degree of synovial effusion, synovial proliferation, osteophytes, subchondral sclerosis, and cartilage quality were scored (Table S3, Additional file 1)). The subchondral bone was divided into 8 sub-regions: second carpal bone (2CB), third carpal bone-radial facet (3CBr), third carpal bone-intermediate facet (3CBi), fourth carpal bone (4CB), ulnar carpal bone (UCB), intermediate carpal bone (ICB), radial carpal bone-third facet (where the fragment was) (RCB3), and radial carpal bone-second facet (RCB2) and each subregion was scored individually (Fig. 2). The size of the fragments was measured, fragment healing was scored, and fragments were classified as mono-articular fragments or bi-articular fractures. Fragment healing was graded as 0 (no evidence of healing), 1 (incomplete healing), or 2 (complete healing). Bone edema-like signal could not be assessed, as fat-suppressed MR images were not available.
Macroscopic pathology
After diagnostic imaging, middle carpal joints were opened by careful sharp dissection and photographed in detail for later blinded assessment of macroscopic pathology.
Macroscopic pathology was scored from detailed photographs by observers blinded to treatment group using a detailed score developed by our group60. In brief, each carpal bone in the middle carpal joint was assessed separately according to erosion severity and extent of erosions, which were multiplied resulting in a total cartilage erosion score (Copenhagen Equine Total Cartilage Score, CEqTCS). The osteochondral fragment of the RCB and the synovial membrane were scored separately.
Tissue sampling
Tissue samples were collected and processed for histology and gene analysis immediately after euthanasia in a clean manner to avoid contamination. Synovial membrane samples were collected using sharp scissors. Osteochondral wedge sections (5 × 15 × 20 mm) were cut from the OA joints using an oscillating saw. Samples were collected from the RCB at the fragment site; the radial facet of the 3CB at the site of the expected kissing lesion (3CBr); the 3CBi, and the ICB. Only osteochondral samples from the 3CBi and ICB were taken from the Control-joints.
Histology
Synovial membrane samples were fixed in formalin for one week, embedded in paraffin, and sectioned at 2 µm thickness and stained with hematoxylin and eosin (HE). Osteochondral samples were fixed in formic acid, decalcified in EDTA, embedded in paraffin, cut into 2 µm sections and stained with HE, or safranin O and fast green (SOFG). All osteochondral samples were sliced in the sagittal plane.
All histology slides were scored by an external assessor. The scoring was done according to recommendation from the OARSI histopathology initiative61. The observer had no knowledge about the study objective and was blinded to study design, treatment groups, and anatomic location. Total cartilage histology scores from the osteochondral sections from the 3CBi and ICB were judged separately for comparison between groups, because lesions in these areas are a consequence of degenerative joint disease rather than direct mechanical injury caused by the surgically created fragment. The kissing lesion (3CBr) and the fragment area (RCB) were also evaluated separately. Images of sections were obtained using an Olympus BX45 microscope, an Olympus UC30 camera and processed with Olympus cellSens Entry 2.1 software.
MSC tracking by PCR of the Y-chromosome SRY
Integrin α10-MSC tracking was performed using an XY-model, where the female horses were treated with male MSCs. PCR analysis was used to detect the presence of a Y-chromosome specific gene (Y-PCR). This was done in the treated group only.
Isolation of DNA
Synovial membrane samples and flakes of cartilage harvested from the 3CBr, 2CB, the ICB and from the fragment site in the RCB of both the OA-joint and the Control-joint. Samples were snap frozen in liquid nitrogen. Subchondral bone debris from the fragment repair site was curetted out and snap frozen in liquid nitrogen. All samples were stored at -80oC until further processing. Cartilage and synovial membrane samples were homogenized in a tissue mill submerged in liquid nitrogen before DNA isolation. DNA was isolated directly from SF cell pellets and subchondral bone debris. DNA was isolated using DNeasy Blood and Tissue kit (Qiagen, USA) according to manufacturer’s protocol and quantified using NanoDrop 2000 spectrophotometer (ThermoFisher Scientific, MA, USA).
PCR analysis
PCR was performed to determine the presence of Y-chromosome specific target gene Sex-determining Region of the Y-chromosome (SRY)62. Forward primer: 5’-CTTAAGCTTCTGCTATGTCCAGAGTATCC-’3. Reverse: 5’-GCGGTTTGTCACTTTTCTGTGGCATCTT-’3.
Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as reference gene; Forward primer: 5’-GGGTGGAGCCAAAAGGGTCATCAT-’3. Reverse: 5’-AGCTTTCTCCAGGCGGCAGGTCAG-’3. Analyses were performed on a 480 LightCycler Instrument (Roche Life Science GmbH, Mannheim, Germany). A positive calibrator sample and a negative control sample were included in all analyses. All analyses were run as triplicates. Samples were determined as positive only when all of the following criteria were achieved; two or more replicate wells were positive, melting curve analysis showed a product melting point consistent with the gene of interest, and subsequent sequencing63 of the PCR product yielded a positive result when blasted against the equine genome.
Statistical Analyses
For repeated measures of numerical values (carpal skin temperature and joint circumference), a mixed effect model was fitted to the data, with horse as a random effect, and with group (untreated or treated) and time (day after MSC treatment) as fixed effects. For repeated measurement of ordinal values (AAEP lameness grade and flexion test scores) a binomial logistic regression model was used with two categories in each (lameness: grade 0–1 versus grade 2–4; flexion test: grade 0 versus ≥ 1).
A Shapiro-Wilk test, histograms and qq-plots were used to assess normality of data. A Student’s t-test was used for pairwise comparison of parametric data (skin temperature, joint circumference) and Wilcoxon Exact rank sum test was used to assess non-parametric or ordinal data (lameness, histopathology, gross pathology, imaging). Multiple Mann-Whitney tests were used to analyze differences between groups for cytokine/chemokine levels and SF cytology at each time point and to analyze differences in cytokines/chemokines in the same group from day 0 onward.
A 5% level of significance was chosen, and raw p-values are shown. All results with a p < 0.07 are mentioned in the results section.