Experimental design and sampling
Six horses, free from pathology affecting the middle carpal or tarsocrural joints, were enrolled in a complete block, randomized crossover study design. The experiment was designed to compare the clinical and synovial fluid biochemical and biophysical changes in response to IL-1β induced synovitis in the middle carpal joint (MCJ) and intra-articular lavage in the tarsocrural joint (TCJ). Breeds included Thoroughbred (2), Mustang/Arabian (2), and Warmblood (2), with ages ranging from 8 to 20 years (mean: 14.5 years, median: 15 years) and weights ranging from 511 to 636 kg (mean: 570 kg, median: 568.2 kg). An equal number of mares and geldings were included. Lameness and physical examinations were conducted prior to the study by two veterinarians to ensure that horses had an AAEP lameness grade of 2/5 or less, and carpal and tarsal radiographs were obtained to rule out any pre-existing joint pathology. Three of the horses had no lameness, two horses had a grade 1/5 lameness, and one horse had a grade 2/5 lameness. The horses all had minimal to no joint effusion in the MCJ or TCJ prior to commencement of the study. Treatment limbs were randomized using a Microsoft Excel random number generator with MCJ and TCJ groups being independent from each other. All horses received IL-1β induced synovitis in the MCJ first, followed by a washout period and the TCJ intra-articular lavage.
One hundred nanograms of recombinant equine IL-1β in 1 mL phosphate buffered saline (PBS) was administered into one randomly assigned middle carpal joint (MCJ), and the contralateral MCJ was injected with 1 mL PBS as a sham/vehicle injection. The recombinant equine IL-1β was all from the same lot number and was aliquoted on arrival to limit freeze thaw cycles. Synovial fluid (SF) and blood were collected at 0, 6, 12, 24, 48, 72, 168 (1 week), 336 (2 weeks), 504 (3 weeks), 672 (4 weeks), and 840 (5 weeks) hours post-induction of synovitis. At each time point, 3 mL of synovial fluid was collected, except the 840 h timepoint when as much fluid as possible was obtained. Following the last collection time point, a synovial membrane biopsy of the MCJ was performed under standing sedation using a small arthrotomy in the dorsolateral aspect of the MCJ and Ferris-Smith rongeurs. These incisions were closed in one layer of non-absorbable simple, interrupted sutures and were kept bandaged for 3 days following suture removal at day 14 post-procedure.
Following a 30-day wash-out period, which consisted of stall rest or small paddock turn-out, the same six horses entered the second stage of the study. This consisted of an intra-articular lavage performed under standing sedation via placement of four 14 g needles into each of the four quadrants of the tarsocrural joint (TCJ)—dorsomedial, dorsolateral, plantaromedial and plantarolateral. Lavage was performed using 2 L of Lactated Ringer’s Solution (LRS) into a randomly selected TCJ. The contralateral limb received no intervention. Synovial fluid and blood were collected over a 35-day period at the same time intervals as described for the synovitis intervention. Following the final collection timepoint at 840 h (5 weeks), a standing synovial membrane biopsy of the TCJ was obtained in four horses, while two horses were euthanized and samples were collected immediately post-mortem. Synovial fluid and blood were obtained ante-mortem.
All work was conducted under approval by the Institutional Animal Care and Use Committee (2018–0024). All joints were clipped and aseptically prepared prior to administration of recombinant IL-1β or intra-articular lavage, and synovial fluid sampling was performed using aseptic technique. One horse required general anesthesia for the intra-articular lavage due to intractable behavior.
Evaluation of clinical response to treatment
A physical examination including heart rate, respiratory rate, rectal temperature, joint circumference, and range of motion, was performed at 0, 6, 12, 24, 48, 72, 168 (1 week), 336 (2 weeks), 504 (3 weeks), 672 (4 weeks), and 840 (5 weeks) hours post-induction of synovitis and lavage. A lameness examination at the walk and trot was performed at each time point, with a veterinarian assessing lameness at the walk and trot and an inertial sensor-based system (Equinosis Q) evaluating gait symmetry at the trot in hand on a firm, synthetic surface. For the synovitis model, bilateral carpal flexions were performed. No flexions were performed for the lavage model.
Joint circumference was measured with a tape measure at three locations for both carpi and tarsi, and the three measurements were averaged to obtain a single measure of effusion/edema. For the MCJ, circumference was measured at the accessory carpal bone (ACB), 2 cm distal to the ACB, and 4 cm distal to the ACB. For the TCJ, circumference was measured at the medial malleolus of the tibia, 1 cm distal to the malleolus, and 4 cm distal to the malleolus.
Synovial fluid was sampled/obtained prior to treatment (0 h), 6, 12, 24, 48, 72, 168 (1 week), 336 (2 weeks), 504 (3 weeks), 672 (4 weeks), and 840 (5 weeks) hours post-induction of synovitis or lavage. Arthrocentesis of 3 mL synovial fluid from either the MCJ or TCJ was performed aseptically following clinical assessment and lameness examination. Horses were sedated using either xylazine (0.2–0.4 mg/kg IV) or detomidine hydrochloride (0.004–0.016 mg/kg IV) combined with acepromazine maleate injection (0.1–0.2 mg/kg IV) and butorphanol tartrate (0.004–0.016 mg/kg IV). One horse required sedation with morphine sulfate (0.08–0.2 mg/kg IV) rather than butorphanol. Synovial fluid was placed in 15-mL conical polypropylene vials on ice and processed within 2 h of collection. A portion of the aspirate was used for analysis of total nucleated cell count and differential lymphocyte, monocyte, and neutrophil counts using an automated cell counter.
The remainder of the aspirate was centrifuged at 4000 x g for 15 min to remove cells and debris. The synovial fluid supernatants were aliquoted and stored at − 80 °C in 1.5 mL Eppendorf tubes. The cell pellet was re-suspended in 500 μL TRIzol Reagent and stored at − 80 °C. Jugular vein blood was collected into glass Vacutainer blood tubes with and without EDTA and centrifuged as above. Serum and plasma were aliquoted into 2 mL Eppendorf tubes and stored at − 80 °C. From the EDTA tubes, the buffy coat was aspirated from the red blood cell pellet and resuspended in 500 μL TRIzol Reagent and stored at − 80 °C. Once all SF samples were collected, they were assessed and scored subjectively for color/hemorrhage.
A plate-based bicinchoninic acid assay (BCA, ThermoFisher Scientific, Waltham, MA) was performed to measure synovial fluid total protein concentrations colorimetrically . Synovial fluid samples diluted in phosphate buffered saline (1:40 dilution) and a series of bovine serum albumin standards were loaded into a 96 well plate (Corning, Corning, NY) and incubated with BCA Working Reagent for 1 h at room temperature. Absorbance was measured on a plate reader at 562 nm (Tecan, Morrisville, NC).
Synovial fluid lubricin (sandwich ELISA)
Synovial fluid lubricin concentration was measured at all time points by a sandwich ELISA using anti-lubricin monoclonal antibody 9G3 (MABT401; EMD Millipore, Darmstadt) and peanut agglutinin (PNA) (Sigma Aldrich, St. Louis, MO) as previously described [48, 71]. Briefly, after 12 h of coating at 4 °C with 10 mg/mL of PNA in 50 mM sodium bicarbonate buffer, pH 9.5, blocking was performed with (PBS) + 3% EIA-grade BSA (Sigma-Aldrich, St. Louis, MO) for 1 h. Equine purified lubricin standard and diluted equine synovial fluid samples (1:1000) were incubated for 1 h, the plate was washed with PBS + 0.1% Tween20, and monoclonal antibody 9G3 (mAbT401 Anti-Lubricin/Prg4 Clone 9G3) was loaded into the plate at 1:2500 dilution for 1 h. Following a wash cycle as above, goat anti-mouse IgG-horseradish peroxidase (EMD Millipore, Darmstadt, Germany) was added to each well at a 1:4000 dilution for 1 h. Washing three times in PBS + 0.1% Tween20, with a final rinse in PBS alone, was performed. TMB reagent was added (Pierce, Rockford, IL), the reaction was stopped with 1 N H2SO4, and absorbance was measured at 450 nm with 540 nm background subtraction. The intra-assay coefficient of variation for the lubricin assay was 11.6%, and the samples reading higher than the upper limit of detection of the assay due to oversaturation were assigned a value of 2000 μg/mL.
Hyaluronic acid quantification – ELISA and gels
Synovial fluid HA concentration was measured at all time points using a commercially available HA ELISA (Hyaluronan DuoSet ELISA, Cat#: DY3614–05, R&D Systems, Minneapolis, MN) . The distribution of HA molecular weights was determined by gel electrophoresis in a similar manner to that described previously . Synovial fluid samples were diluted 1:15 with phosphate buffered saline and incubated overnight with 75 μg/ mL proteinase k (Proteinase K, recombinant, PCR grade, Roche Applied Science, Mannheim, Germany). Samples and standards, HiLadder (0.5–1.5 MDa) and Mega-HA Ladder (1.5–6.1 MDa; AMS Biotechnology Limited, Cambridge, MA) were loaded onto a 0.5% agarose gel and run at 57 V for 8 h. Gels were stained for 24 h in 0.005% Stains-All (Sigma-Aldrich, St. Louis, MO) in 50% ethanol and de-stained in 10% ethanol for 24 h with final destaining occurring on exposure to ambient light. Images of gels acquired using a Bio-Rad VersaDoc Imaging System (Hercules, CA) with relative band intensity calculated using Fiji Software (ImageJ). The intra-assay coefficient of variation for the HA ELISA was 9.1%.
At day 35 post-induction of synovitis or post-lavage, a 2–3 mm sample of synovial membrane was obtained using a Ferris-Smith rongeur via standing arthrotomy from either the MCJ or the TCJ joint. Sections were fixed in 10% formaldehyde for a minimum of 3 days, dehydrated in alcohol, cleared in xylene, paraffin embedded and sectioned at 6 μm . Slides were stained as one batch with haemotoxylin and eosin for basic cell identification, then evaluated by three blinded assessors.
Chemokine multiplex assay
The equine chemokine multiplex assay has been validated and is performed at the Animal Health Diagnostic Center at Cornell University. The fluorescent bead-based assay simultaneously quantifies six equine cytokines/chemokines (IL-1β, TNF-α, CCL2, CCL3, CCL5, and CCL11) using pairs of monoclonal antibodies (mAbs) for detection of each equine chemokine. The procedures of coupling mAbs to the fluorescent beads (Luminex Corp., Austin, TX, USA) and performing the different steps of the assay were previously described in detail for other equine cytokines  and were identical for this assay. In brief, the following beads were coupled to mAbs: bead 33 with equine TNF-α mAb 292–1, bead 34 with equine CCL11 mAb 24, bead 35 with IL-1β mAb 84–2, bead 36 with CCL5 mAb 91–1, bead 37 with CCL2 mAb 104–2, and bead 42 with CCL3 mAb 77–2. Specificity to respective chemokine and recognition of the native proteins were confirmed for all mAbs  before they were used in the multiplex assay.
All six recombinant equine proteins were expressed in mammalian cells as IL-4 fusion proteins [43, 76]. For the assay runs, a mixture of the six recombinant chemokines was included in different concentrations (5-fold dilutions in PBS with 1% (w/v) BSA and 0.05% (w/v) sodium azide (blocking buffer)) to create standard curves for quantification of all six chemokines in equine samples. Joint samples were diluted 1:2 in blocking buffer. Millipore Multiscreen HTS plates (Millipore, Danvers, MA) were soaked with PBS with 0.1% (w/v) BSA, 0.02% (v/v) Tween 20 and 0.05% (w/v) sodium azide (PBS-T) using a ELx50 plate washer (Biotek Instruments Inc., Winooski, VT) for 2 min. The solution was aspirated from the plates and 50 μl of each diluted standard dilution or the samples were applied to the plates. Then, 50 μl of bead solution, containing 5 × 103 beads per bead number, was added to each plate well and incubated with the standards or samples for 30 min on a shaker at room temperature. The plates were washed with PBS-T and 50 μl of the equine detection antibody mixture diluted in blocking buffer was added to each well and incubated for 30 min as above. The detection antibody mixtures included six biotinylated mAbs: TNF-α mAb 48–1, CCL11 mAb 25, IL-1β mAb 62–7, CCL5 mAb 46–1, CCL2 mAb 49, and CCL3 mAb 289–2 . Afterwards plates were washed again and 50 μl of streptavidin-phycoerythrin (Invitrogen, Carlsbad, CA) was added to the plates for another 30 min incubation as above. Plates were washed for a last time, beads were resuspended in 100 μl of blocking buffer, and the plates were placed on the shaker for 15 min. The assay was analyzed in a Luminex 200 instrument (Luminex Corp., Austin, TX, USA). The data were reported as median fluorescent intensities. For standard curve fitting and subsequent calculation of the chemokine concentrations in samples the logistic 5p formula (y = a + b/(1 + (x/c)ˆd)ˆf) was used (Luminex 200 Integrated System). Chemokine concentrations were reported in pg/ml. For TNF-α and IL-1β the non-detectable values (0) were set to 1 for analysis.
Dimethylmethylene blue (DMMB) assay
Synovial fluid samples were tested at all time points for sulfated glycosaminoglycan (sGAG) concentration using a 1,9-dimethylmethylene blue (DMMB) assay. DMMB dye was prepared by combining 16 mg DMMB dye (Sigma-Aldrich, St. Louis, MO) with 5 mL 95% ethanol and incubating at room temperature for 30 min. Two milliliters of pH 3.5 formate buffer was added to solution, and the total volume was adjusted to 1 L with water. The dye was stored at room temperature, protected from light. Synovial fluid samples were incubated at 37 °C with 30 U/mL Streptomyces hyaluronidase (Sigma-Aldrich, St. Louis, MO) for 1 h and were vortexed every 20 min during the digestion period. Digested samples were diluted to a final concentration of 1:30 in water. A solution of chondroitin 4-sulfate (Sigma-Aldrich, St. Louis, MO) was used as a standard in a range of concentrations from 2.5 μg/mL to 30 μg/mL. Milli-Q water was used as a blank. All samples, standards, and blanks were plated in duplicate at a volume of 50 μL on a transparent 96-well plate (Corning Inc., Corning, NY). After all samples were applied, the plate was agitated for 60 s on an orbital shaker (Bellco Glass Inc., Vineland, NJ). Two hundred microliters of DMMB dye was then added to each well of the plate using a multi-channel pipette, and absorbance was immediately read at 540 nm using a Spark 10 M plate reader (Tecan Austria GmbH, Grödig, Austria).
Synovial fluid concentration of PGE2 was evaluated at all time points as previously described [30, 32]. In brief, 250 μL of synovial fluid (SF) was mixed with 250 μL of 80% ethanol and 5 μL of glacial acetic acid. After incubation, for 5 min at room temperature, and centrifugation (6000 rpm, 8 min), the supernatant was loaded onto Ethyl C2 mini-columns (Agilent Technologies, Santa Clara, CA) that had been equilibrated with 10% ethanol. The PGE2-containing mini-columns were washed with MQ-H2O and hexane sequentially. The PGE2 was eluted with two replicates of 375 μL of ethyl acetate. The combined 750 μL eluate was dried in a Speed Vacuum (Speed Vac Plus, SC110A, SAVANT) and the PGE2 powder was stored in at -70 °C. The PGE2 concentration in SF samples were measured with a highly sensitive and competitive PGE2 ELISA kit (Enzo Life Sciences, Inc., Farmingdale, NY). The PGE2 powder derived from 250 μL of synovial fluid was resuspended in 250 μL of PGE2 assay buffer. Then, 100 μL of the PGE2 solution was added in duplicate to the goat anti-mouse IgG microtiter plate. After being bound with the kit conjugate/antibody and wash, the plate was read at 405 nm with a background reading at 570 nm using SPARK 10 M microplate reader (TECAN, Zürich, Switzerland). The PGE2 levels in SF were evaluated using 4-parameter standard curve with the x-axis at Log scale. If the absorbance reading at 405 nm was above the standard range (maximum 2500 pg/mL), the PGE2 concentration was calculated as the maximum standard value (2500 pg/mL). The intra-assay coefficient of variation for the PGE2 ELISA was 7%.
Synovial fluid viscosity was measured with particle tracking microrheology. 0.5 μm yellow-green fluorescent beads (FluoSpheres™ Carboxylate-Modified Microspheres, 0.5 μm, yellow-green fluorescent) were diluted in a 1:50 ratio with water. The diluted beads were then mixed with synovial fluid in a 1:50 ratio, totaling 20 μl. Samples were loaded into wells of silicone gaskets (Grace Bio-Labs Press-To-Seal silicone isolator, No PSA 24–2 mm diam. × 0.5 mm depth) which were press-sealed on 35 mm glass-bottom dishes (Cellvis D35–20-1.5-N). The sample was covered with a #1.5 glass coverslip to prevent evaporation and placed on an inverted fluorescence microscope (IX81, Olympus) equipped with a 60x NA 1.2 water-immersion objective and with 1x magnification. Fluorescence excitation by a 488 nm laser (Sapphire-LP, Coherent) was expanded 8.3x before focusing on the objective back aperture by a 300 mm tube lens (ThorLabs). Fluorescence emission was imaged with an EMCCD (897 Ultra, Andor) through a standard FITC filter set (Chroma) using the Micro-Manager software package (Open Imaging). Three 30-s videos were taken per sample, with several locations in each sample targeted to collect particle movement data. Each video contained approximately 15 particles in the frame. Analysis of motion was done using the Trackpy Python package. Images were acquired at 16 Hz.
RNA extraction/gene expression
Gene expression was examined in synovial membrane (SM) samples of horses (Equus caballus) collected at day 35 post-induction of synovitis or intra-articular lavage and in synovial fluid cell pellets collected at 12- and 24 h post-induction. RNA was extracted from synovial membrane samples using the RNeasy Lipid Tissue Mini kit (QIAGEN, Gaithersburg, MD). The synovial fluid cell pellets were centrifuged as described above, the supernatant was aliquoted for freezing, and the SF cell pellet was suspended in 0.5 mL of TRIzol Reagent (ThermoFisher Sci., Waltham, MA). Crude RNA was extracted following the instruction manual of TRIzol Reagent. An RNA Clean and Concentrator kit (Zymo Research, Irvine, CA) was employed to further purify and concentrate the SF cell pellet RNA. Any remaining genomic DNA in the RNA extract was removed by DNase I digestion on-column for both SM and SF cell pellet RNA. RNA concentrations and quality were determined using a 16-well NanoQuant plate and a SPARK 10 M microplate reader (TECAN, Zürich, Switzerland). The expression levels of three genes (PRG4, IL1β and HAS2 encoding for Hyaluronan synthase 2) in SM and four genes (PRG4, TSG6 encoding for TNF-stimulated gene 6 protein, HAS1 and HAS3) in SF cell pellets were analyzed.
Gene expression was detected by quantitative real-time PCR (qRT-PCR) using the Applied Biosystems Real-Time PCR ViiA 7 system (Applied Biosystems, Foster City, CA). All samples were analyzed in duplicate using the Power SYBR green RNA-to-CT one-step kit (Applied Biosystem Inc., Carlsbad, CA). Primers (Supplemental Data 3) were selected from publications [48, 59] or designed by NCBI Primer 3 & Blast or with DNASTAR LASERGENE.
For qRT-PCR, 30 ng of SM total RNA or 15 ng of SF cell pellet RNA was used in 20 μL of reaction mix containing SYBR RT-PCR mix and RT enzyme mix. The qRT-PCR was run at 48 °C for 30 min and at 95 °C for 10 min, followed by 40 cycles of 95 °C/15 s and 60 °C/1 min. Successful qRT-PCR was verified via analysis of both dissociation curves and agarose gel electrophoresis. All values were normalized to the housekeeping gene 18S rRNA. Relative gene expression was analyzed using the 2-∆CT method [77, 78], where ∆CT = CT (gene of interest)-CT (18S rRNA) and calculated as (104–107) × 2-∆CT .
All analyses were conducted with Stata 16.1MP, StataCorp, College Station TX, with two-sided tests of hypotheses and a p-value < 0.05 as the criterion for statistical significance. Descriptive analyses include computation of means (with 95% confidence intervals [95%CI]), standard deviations, medians, interquartile ranges (IQR) of continuous variables and tabulation of categorical variables. Tests of normal distribution (Shapiro-Wilk test) were performed to determine extent of skewness. Frequency counts and percentages were used for categorical variables such as signalment and others.
Inference statistical analysis was based on a multilevel mixed-effects model with interaction between treatment group and categorical time as the fixed effects and age as a confounder. Random effects were set on the level of joint nested within leg which in term was nested within specific animal. All random effects were considered random intercepts. Robust estimation of the variance was used to permit for departures from normality of the outcome. Post-hoc pairwise comparisons were conducted to estimate the marginal (model adjusted) effects. Least significant difference (LSD) was used to adjust for multiple comparisons. The figures presented in this paper depict the marginal means used in the statistical model and not the raw data which is presented in Supplemental Data 2.